Voted #1 \"Best Places to Work\" by the Phoenix Business Journal, Achieve is a leader in the digital personal finance industry built to help everyday people thrive

Achieve

achieve

Authors:
(1) Simone Silvestri, Massachusetts Institute of Technology, Cambridge, MA, USA;
(2) Gregory Wagner, Massachusetts Institute of Technology, Cambridge, MA, USA;
(3) Christopher Hill, Massachusetts Institute of Technology, Cambridge, MA, USA;
(4) Matin Raayai Ardakani, Northeastern University, Boston, MA, USA;
(5) Johannes Blaschke, Lawrence Berkeley National Laboratory, Berkeley, CA, USA;
(6) Valentin Churavy, Massachusetts Institute of Technology, Cambridge, MA, USA;
(7) Jean-Michel Campin, Massachusetts Institute of Technology, Cambridge, MA, USA;
(8) Navid Constantinou, Australian National University, Canberra, ACT, Australia;
(9) Alan Edelman, Massachusetts Institute of Technology, Cambridge, MA, USA;
(10) John Marshall, Massachusetts Institute of Technology, Cambridge, MA, USA;
(11) Ali Ramadhan, Massachusetts Institute of Technology, Cambridge, MA, USA;
(12) Andre Souza, Massachusetts Institute of Technology, Cambridge, MA, USA;
(13) Raffaele Ferrari, Massachusetts Institute of Technology, Cambridge, MA, USA. Authors: Authors: (1) Simone Silvestri, Massachusetts Institute of Technology, Cambridge, MA, USA; (2) Gregory Wagner, Massachusetts Institute of Technology, Cambridge, MA, USA; (3) Christopher Hill, Massachusetts Institute of Technology, Cambridge, MA, USA; (4) Matin Raayai Ardakani, Northeastern University, Boston, MA, USA; (5) Johannes Blaschke, Lawrence Berkeley National Laboratory, Berkeley, CA, USA; (6) Valentin Churavy, Massachusetts Institute of Technology, Cambridge, MA, USA; (7) Jean-Michel Campin, Massachusetts Institute of Technology, Cambridge, MA, USA; (8) Navid Constantinou, Australian National University, Canberra, ACT, Australia; (9) Alan Edelman, Massachusetts Institute of Technology, Cambridge, MA, USA; (10) John Marshall, Massachusetts Institute of Technology, Cambridge, MA, USA; (11) Ali Ramadhan, Massachusetts Institute of Technology, Cambridge, MA, USA; (12) Andre Souza, Massachusetts Institute of Technology, Cambridge, MA, USA; (13) Raffaele Ferrari, Massachusetts Institute of Technology, Cambridge, MA, USA. Table of Links Abstract and 1 Justification Abstract and 1 Justification 2 Performance Attributes 2 Performance Attributes 3 Overview of the Problem 3 Overview of the Problem 4 Current State of the Art 4 Current State of the Art 5 Innovations 5 Innovations 5.1 Starting from scratch with Julia 5.1 Starting from scratch with Julia 5.2 New numerical methods for finite volume fluid dynamics on the sphere 5.2 New numerical methods for finite volume fluid dynamics on the sphere 5.3 Optimization of ocean free surface dynamics for unprecedented GPU scalability 5.3 Optimization of ocean free surface dynamics for unprecedented GPU scalability 6 How performance was measured 6 How performance was measured 7 Performance Results and 7.1 Scaling Results 7 Performance Results and 7.1 Scaling Results 7.2 Energy efficiency 7.2 Energy efficiency 8 Implications 8 Implications 9 Acknowledgments and References 9 Acknowledgments and References 6 How performance was measured The Oceananigans model performance is estimated for two near-global ocean simulations with different domains: a realistic (R) domain and an aqua planet (AP) domain. Both domains span the entire longitudinal extent of the sphere and cover a latitude range of 75◦S to 75◦N. The Realistic domain has realistic bathymetry and is forced by realistic surface momentum, heat, and salinity fluxes derived from the ECCO2 state estimate[31] at three resolutions: • OceananigansR12 with 1/12th degree horizontal resolution (∼7 km) and 48 vertical levels OceananigansR12 • OceananigansR24 with 1/24th degree horizontal resolution (∼3.4 km) and 100 vertical levels OceananigansR24 • OceananigansR48 with 1/48th degree horizontal resolution (∼1.7 km) and 100 vertical levels OceananigansR48 Figure 3 shows surface vertical vorticity after one year integration of OceananigansR12 and OceananigansR48 over the global ocean and also for selected regions to show further detail. Both OceananigansR48 and OceananigansR12 exhibit macroscale turbulent ocean features that are currently unresolved by most IPCC-class models. The OceananigansR48 solution exhibits fronts, filaments, and other “submesoscale” vorticity features realized only a handful of times in global simulations. OceananigansR12 OceananigansR48 OceananigansR12 OceananigansR48 The idealized OceananigansAP suite of simulations [13], which has idealized bathymetry and surface forcing that does not require interpolation to different resolutions, is used for weak scaling experiments. All OceananigansAP experiments have 100 vertical levels and two latitudinal ridges that divide the world ocean into two basins. We vary the horizontal resolution of OceananigansAP from 1/6th of a degree (∼14 km) to 1/196th of a degree (∼488 m). OceananigansAP OceananigansAP OceananigansAP None of our simulations require explicit horizontal diffusion of momentum or tracers owing to the adaptive WENO advection scheme described in section 5.2. All simulations use a Richardson-number based parameterization for vertical mixing due to unresolved shear and convective turbulence at 1–100 m scales. To assess the time-to-solution for each experiment in simulated years per day (SYPD), we measure average wall clock time per time-step. Wall clock time is sampled through NVIDIA’s Nsight System and recorded by NVIDIA Tool Extension Library via the NVTX.jl Julia package. To assess the efficiency of each solution in simulated years per mega-watt-hour (SYPMWh), we combine SYPD with an estimate of the mean power draw over the duration of an experiment. On MIT Satori [2], which has 256 Nvidia V100s, we have access to precise, billing-grade power metering. For all simulations with Nvidia A100s we estimate power consumption P with where D is the number of A100s and N is the number of nodes. We further note that power estimates are provided by LICOM3 and COSMO, but not for LLC4320 or Veros. To estimate the power consumption of LLC4320, we assume that each of the 1000 dual CPU nodes draws 500W. We estimate the power consumption of iHESP CESM [51] and HadGCM3 [37] as a percentage of the peak power consumption of their respective clusters. We use equation (1) to estimate Veros’ power consumption on 1 node with 16 A100s. 7 Performance Results We report both scaling results via time-to-solution in SYPD and efficiency results via energy-to-solution in SYPMWh. 7.1 Scaling Results Realistic ocean simulations (Satori and Engaging clusters). We report strong scaling tests using the realistic global setup shown in figure 3 on two clusters: (i) the MIT Satori cluster [2], a high-performance Power 9 system composed of 64 Power 9 nodes hosting four Nvidia V100 GPUs with 32GBs memory each, and (ii) the Engaging MIT cluster, using 8 nodes that host 4 NVlinked A100s with 80GBs memory each. The resulting wall clock time per time step, averaged over 1500 time steps, is presented in Figure 4 for both single precision (FP32) and double precision (FP64) computations. On a single node, OceananigansR12 attains 0.9 SYPD in double precision and 1.4 SYPD in single precision, with a wall clock time per time step ranging from 330 to 550 ms. When increasing the number of nodes up to 16 (64 GPUs), the communication overhead increases, resulting in 12.4 SYPD in single precision and 7.75 SYPD in double precision. We measure a strong scaling efficiency of 52% in single precision and 55% in double precision over 64 GPUs, because the computational workload (40 ms wall clock time per time-step) eventually becomes too short to completely mask the communication overhead. Realistic ocean simulations (Satori and Engaging clusters). OceananigansR12 For higher-resolution ocean weather-permitting simulations, the scaling is almost ideal across the range we investigate. For OceananigansR24 (FP64-V100) and OceananigansR48 (FP32-V100), we measure larger than ideal scaling. This counter-intuitive result is a product of a load balance improvement as the number of GPUs increases. In summary, we attain 1.94 SYPD on 120 V100 GPUs with a kilometer-scale resolution (OceananigansR24) and 0.33 SYPD with an ocean weather resolving simulation (OceananigansR48). Finally, we have tested the OceananigansR48 setup on 144 Perlmutter nodes (576 A100 GPUs), reaching the 0.95 SYPD. This is the first instance of a kilometer-scale ocean achieving ∼1 SYPD. We have also tested the OceananigansR12 setup on 17 nodes obtaining 9.9 SYPD (see fig. 5). OceananigansR24 OceananigansR48 OceananigansR24 OceananigansR48 OceananigansR48 Aqua-planet simulation (Perlmutter cluster). We report weak scaling tests on the NERSC supercomputer (Perlmutter). Perlmutter is a HPE (Hewlett Packard Enterprise) Cray EX super Aqua-planet simulation (Perlmutter cluster). computer that hosts four A100 GPUs with 40GB per node, linked through a NVLink3 interconnect. All weak scaling tests are performed using the OceananigansAP setup on double precision. We allocate two different horizontal resolutions (1/12 and 1/6 of a degree), progressively increasing them with the number of GPUs while maintaining 100 vertical levels. As shown in figure 5, we obtain 100% weak scaling efficiency for the whole investigated range (1 to 196 nodes – 4 to 768 A100s). OceananigansAP This paper is available on arxiv under CC BY 4.0 DEED license. This paper is available on arxiv under CC BY 4.0 DEED license. available on arxiv

Scaling the Depths: Oceananigans Achieves Record-Breaking Climate Simulation Performance

How can blockchain support confidential transactions while meeting regulatory demands? COTI launches its mainnet as a scalable Layer 2 to support confidential operations in Web3. COTI COTI announces the launch of its mainnet after more than two years of research and development. The network operates as a Layer 2 solution for confidential transactions and aims to serve as a bridge between privacy and compliance in the digital ecosystem. The platform utilizes an implementation of Garbled Circuits developed in collaboration with Soda Labs. This system is designed to offer significant speed improvements over existing solutions while providing a Privacy-on-Demand service across multiple blockchain networks. Addressing the Privacy Gap in Web3 The absence of viable privacy layers has limited institutional participation in areas such as Real World Assets (RWAs). This gap has prevented many institutions from protecting sensitive data and meeting compliance requirements. COTI's new mainnet is designed to enable businesses to secure proprietary data and allow individuals to access Web3 services without disclosing personal information. The system also supports AI agents to operate as independent entities, ensuring that both data privacy and operational compliance are maintained across various use cases. Technical Framework and Ecosystem Integration COTI's mainnet employs a Layer 2 architecture that records transactions on-chain while supporting high-speed processing. The use of Garbled Circuits enhances the speed of confidential transactions by up to 3,000 times compared to traditional methods. The network is structured to support Privacy-on-Demand, providing confidentiality while meeting compliance standards. A range of ecosystem partners has already integrated COTI’s privacy layer. Partners include Bancor with its Carbon DeFi platform, Band Protocol, MyEtherWallet (MEW), and PriveX. In addition, COTI is extending its operations to government entities and CBDC initiatives, including a pilot project with the Bank of Israel involving the Digital Shekel. Further expansion announcements are expected in the near future. Transition and User Incentives To support the mainnet launch, the COTI Foundation will distribute 100 $COTI tokens to tens of thousands of wallets. Users who add COTI’s Layer 2 network to MetaMask will be eligible for a token claim. The release also includes a new Explorer and Bridge to facilitate the upgrade from COTI V1 to V2, ensuring a smooth transition for current users. Final Thoughts The launch of COTI's mainnet signals a shift in how confidential transactions are managed in Web3. By providing a scalable and compliant solution, COTI positions its network as an enabler of secure and private blockchain operations. The move may pave the way for increased adoption in sectors that require both confidentiality and adherence to regulatory standards, potentially reshaping the landscape for digital financial services. Don’t forget to like and share the story! Vested Interest Disclosure: This author is an independent contributor publishing via our business blogging program. HackerNoon has reviewed the report for quality, but the claims herein belong to the author. #DYO Vested Interest Disclosure: This author is an independent contributor publishing via our business blogging program. HackerNoon has reviewed the report for quality, but the claims herein belong to the author. #DYO Vested Interest Disclosure: business blogging program business blogging program

Can Web3 Achieve Confidentiality and Compliance? COTI’s Mainnet Provides a Path

¿Web3 ukax Confidencialidad ukat Cumplimiento ukanak phuqhaspati? COTI’s Mainnet ukax mä Thakhi churaraki

Může Web3 dosáhnout důvěrnosti a souladu? Hlavní síť COTI poskytuje cestu

¿Puede la Web3 lograr confidencialidad y cumplimiento normativo? La red principal de COTI ofrece una solución.

Bisakah Web3 Mencapai Kerahasiaan dan Kepatuhan? Mainnet COTI Menyediakan Jalan

Web3は機密性とコンプライアンスを実現できるか？COTIのメインネットが道を開く

Ar „Web3“ gali užtikrinti konfidencialumą ir atitiktį? COTI pagrindinis tinklas suteikia kelią

Kan Web3 vertrouwelijkheid en naleving bereiken? Het hoofdnetwerk van COTI biedt een pad

Ko Web3 Inogona Kuwana Kuvanzika uye Kutevedzera? COTI's Mainnet Inopa Nzira

Може ли Веб3 да постигне поверљивост и усклађеност? ЦОТИ Маиннет пружа пут

Web3 ரகசியத்தன்மை மற்றும் இணக்கத்தை அடைய முடியுமா? COTI இன் Mainnet ஒரு பாதையை வழங்குகிறது.

Оё Web3 метавонад ба махфият ва мувофиқат ноил шавад? Шабакаи асосии COTI роҳро таъмин мекунад

Xana Web3 Yi Nga Fikelela Xihundla Na Ku Landzelela milawu? COTI’s Mainnet Yi Nyika Ndlela

You Don't Need Deep Pockets to Achieve Impressive Load Times

Unreasonable Goals Are Actually Easier to Achieve: Here's Why

ምክንያታዊ ያልሆኑ ግቦችን ለማሳካት በጣም ቀላል ናቸው፡ ምክንያቱ ይህ ነው።

Неабгрунтаваных мэтаў насамрэч лягчэй дасягнуць: вось чаму

অযৌক্তিক লক্ষ্যগুলি অর্জন করা আসলে সহজ: কেন তা এখানে

Las metas poco razonables son en realidad más fáciles de alcanzar: aquí está el porqué

Os obxectivos non razoables son realmente máis fáciles de acadar: aquí tes por que

Az ésszerűtlen célokat valójában könnyebb elérni: Íme, miért

Gli obiettivi irragionevoli sono in realtà più facili da raggiungere: ecco perché

非現実的な目標は実は達成しやすい：その理由はここにある

غیر معقول اهداف په حقیقت کې ترلاسه کول اسانه دي: دلته ولې

Необоснованные цели на самом деле легче достичь: вот почему

Zvinangwa Zvisinganzwisisike Zviri Nyore Kuzadzisa: Heino Chikonzero

ዘይምኽንያታዊ ዕላማታት ብሓቂ ንምዕዋት ዝቐለለ እዩ፦ ስለምንታይ ከምኡ ከም ዝኾነ እንሆ

Global Recognition Awards is adding new finance-focused categories, reflecting the increasing demand for acknowledgment in the field. With banking, fintech, and sustainability efforts gaining traction, more organizations are being recognized for their achievements. Finance Awards Becoming More Common Recognition in finance is happening more often. Organizations like Global Finance Magazine, FinanceAsia, and the Asian Banker continue to spotlight achievements in areas such as retail banking and sustainable finance. This interest in finance awards points to a wider effort to acknowledge leadership and progress. To align with this, the Global Recognition Awards has introduced new categories to honor professionals and companies that are making an impact. "We've received more requests to acknowledge those shaping finance," says Jethro Sparks, CEO of Global Recognition Awards. "These categories celebrate individuals and businesses leading the way in digital banking, sustainability, and financial services." The newly added categories include fintech advancements, responsible finance, digital banking, and accomplishments in retail and corporate banking. With these updates, the Global Recognition Awards gives a wider range of financial professionals and institutions the chance to gain visibility. Local Achievements With International Exposure Unlike other awards that focus on specific regions, Global Recognition Awards takes a worldwide perspective. It welcomes both major corporations and smaller financial institutions that want their contributions acknowledged. "Financial developments don’t stay confined to one place—they expand quickly," Sparks explains. "Our program connects local efforts with audiences across different regions, giving professionals in finance more awards to win on an international stage." For financial institutions in developing areas, this kind of exposure can be transformative. Smaller firms, often overshadowed by larger players, benefit from inclusion in a widely respected program. Recognition provides credibility and new possibilities. Strong Focus on Responsible Finance and Technology Sustainability is becoming a greater priority in finance, and the Global Recognition Awards is making room for this adjustment. New categories acknowledge leaders in green banking, responsible finance, and ESG (environmental, social, and governance) practices. "Finance influences the direction of sustainability," Sparks says. "Recognizing these efforts encourages more organizations to integrate responsibility into their strategies." Advances in technology are also changing financial services. The awards now include categories for blockchain, artificial intelligence in finance, and digital payments. These additions highlight major developments and the tools shaping the next era of finance. Why Recognition Matters in Finance With competition increasing, being acknowledged by a respected program strengthens a financial institution’s reputation. Winning a Global Recognition Award helps businesses attract clients, improve their public image, and gain visibility in their field. For smaller firms, this can be a turning point. "We've seen winners gain new partnerships, attract top professionals, and catch the attention of investors," Sparks says. "Trust is essential in finance, and outside recognition reinforces that trust." A careful selection process guarantees that recipients meet high standards, making the award meaningful. Institutions can reference these achievements in marketing, investor materials, and public relations to reinforce their credibility. Finance continues to change, and awards like the Global Recognition Awards provide a way to celebrate success and acknowledge progress in the field. Keeping up with updates and expanding recognition efforts allows these awards to reflect industry advancements. Global Recognition Awards is adding new finance-focused categories, reflecting the increasing demand for acknowledgment in the field. With banking, fintech, and sustainability efforts gaining traction, more organizations are being recognized for their achievements. Finance Awards Becoming More Common Finance Awards Becoming More Common Recognition in finance is happening more often. Organizations like Global Finance Magazine, FinanceAsia, and the Asian Banker continue to spotlight achievements in areas such as retail banking and sustainable finance. This interest in finance awards points to a wider effort to acknowledge leadership and progress. finance awards finance awards To align with this, the Global Recognition Awards has introduced new categories to honor professionals and companies that are making an impact. "We've received more requests to acknowledge those shaping finance," says Jethro Sparks, CEO of Global Recognition Awards. "These categories celebrate individuals and businesses leading the way in digital banking, sustainability, and financial services." "We've received more requests to acknowledge those shaping finance," "These categories celebrate individuals and businesses leading the way in digital banking, sustainability, and financial services." The newly added categories include fintech advancements, responsible finance, digital banking, and accomplishments in retail and corporate banking. With these updates, the Global Recognition Awards gives a wider range of financial professionals and institutions the chance to gain visibility. Local Achievements With International Exposure Unlike other awards that focus on specific regions, Global Recognition Awards takes a worldwide perspective. It welcomes both major corporations and smaller financial institutions that want their contributions acknowledged. "Financial developments don’t stay confined to one place—they expand quickly," Sparks explains. "Our program connects local efforts with audiences across different regions, giving professionals in finance more awards to win on an international stage." "Financial developments don’t stay confined to one place—they expand quickly," "Our program connects local efforts with audiences across different regions, giving professionals in finance more awards to win on an international stage." awards to win awards to win For financial institutions in developing areas, this kind of exposure can be transformative. Smaller firms, often overshadowed by larger players, benefit from inclusion in a widely respected program. Recognition provides credibility and new possibilities. Strong Focus on Responsible Finance and Technology Strong Focus on Responsible Finance and Technology Sustainability is becoming a greater priority in finance, and the Global Recognition Awards is making room for this adjustment. New categories acknowledge leaders in green banking, responsible finance, and ESG (environmental, social, and governance) practices. "Finance influences the direction of sustainability," Sparks says. "Recognizing these efforts encourages more organizations to integrate responsibility into their strategies." "Finance influences the direction of sustainability," "Recognizing these efforts encourages more organizations to integrate responsibility into their strategies." Advances in technology are also changing financial services. The awards now include categories for blockchain, artificial intelligence in finance, and digital payments. These additions highlight major developments and the tools shaping the next era of finance. Why Recognition Matters in Finance With competition increasing, being acknowledged by a respected program strengthens a financial institution’s reputation. Winning a Global Recognition Award helps businesses attract clients, improve their public image, and gain visibility in their field. For smaller firms, this can be a turning point. "We've seen winners gain new partnerships, attract top professionals, and catch the attention of investors," Sparks says. "Trust is essential in finance, and outside recognition reinforces that trust." "We've seen winners gain new partnerships, attract top professionals, and catch the attention of investors," "Trust is essential in finance, and outside recognition reinforces that trust." A careful selection process guarantees that recipients meet high standards, making the award meaningful. Institutions can reference these achievements in marketing, investor materials, and public relations to reinforce their credibility. Finance continues to change, and awards like the Global Recognition Awards provide a way to celebrate success and acknowledge progress in the field. Keeping up with updates and expanding recognition efforts allows these awards to reflect industry advancements.

Global Recognition Awards Shakes Up Finance Industry With New Categories for Top Achievers

Узнагароды Global Recognition Awards ускалыхнулі фінансавую індустрыю новымі катэгорыямі для лепшых дасягненняў

Наградите Global Recognition Awards разтърсват финансовата индустрия с нови категории за най-добри постижения

শীর্ষস্থানীয় অর্জনকারীদের জন্য নতুন বিভাগগুলির মাধ্যমে বিশ্বব্যাপী স্বীকৃতি পুরষ্কারগুলি অর্থ শিল্পকে নাড়া দিয়েছে

Los premios Global Recognition Awards revolucionan la industria financiera con nuevas categorías para los mejores ganadores

Globaalit tunnustuspalkinnot ravistelevat rahoitusalaa uusilla huippusuorituskategorioilla

ग्लोबल रिकॉग्निशन अवार्ड्स ने शीर्ष उपलब्धि प्राप्तकर्ताओं के लिए नई श्रेणियों के साथ वित्त उद्योग में हलचल मचा दी है

Penghargaan Global Recognition Mengguncang Industri Keuangan dengan Kategori Baru untuk Peraih Prestasi Terbaik

グローバル・レコグニション・アワードがトップ達成者のための新カテゴリーで金融業界を揺るがす

Жаһандық мойындау марапаттары үздік жетістіктерге арналған жаңа санаттармен қаржы саласын дүр сілкіндіреді

글로벌 인정상, 최고 성취자를 위한 새로운 부문으로 금융 산업을 뒤흔들다

Ba Prix Ya Reconnaissance Mondiale Eningisi Industrie Finances Na Ba Catégories Ya Sika pona ba Top Achievers

รางวัล Global Recognition สร้างความสั่นสะเทือนให้กับอุตสาหกรรมการเงินด้วยหมวดหมู่ใหม่สำหรับผู้ประสบความสำเร็จสูงสุด

Table of Links Abstract and 1 Introduction 2 Related Works 3 Method and 3.1 Proxy-Guided 3D Conditioning for Diffusion 3.2 Interactive Generation Workflow and 3.3 Volume Conditioned Reconstruction 4 Experiment and 4.1 Comparison on Proxy-based and Image-based 3D Generation 4.2 Comparison on Controllable 3D Object Generation, 4.3 Interactive Generation with Part Editing & 4.4 Ablation Studies 5 Conclusions, Acknowledgments, and References SUPPLEMENTARY MATERIAL A. Implementation Details B. More Discussions C. More Experiments 4.2 Comparison on Controllable 3D Object Generation We compare our method with Controllable 3D object generation methods, including Latent-NeRF [Metzer et al. 2023] and Fantasia3D [Chen et al. 2023a]. Latent-NeRF introduces a sketch-shape guided loss, which constrains the density field close to the surface of the shape proxy, while Fantasia3D uses coarse shapes as the geometry initialization of DMTet [Shen et al. 2021]. In the experiment, we give all the methods the same coarse shape and the text prompts as a guidance and output the neural reconstructions’ rendered views for comparison (since Latent-NeRF does not officially provide mesh extraction). As shown in Fig. 4, Latent-NeRF generally obtains plausible results but fails to produce tiny shape and appearance details (e.g., blurry textured sofa, fox missing clear eyes and right arms and missing sunflower in Fig. 5 (b)), which indicates that directly applying 3D control to the 3D representation is sub-optimal since it might not work smoothly with the generic SDS loss. For Fantasia3D, since it only uses the 3D shape for initialization rather than supervision, it often generates overgrowth results that do not follow the given shape (e.g., inflate sofa, the fox with incomplete downward arms, pot with many leaves and a broken sunflower in Fig. 5 (c)) and also slightly suffers from “multi-face Janus problem” (e.g., multi-face fox in Fig. 5 (c)). Since both Latent-NeRF and Fantasia3D use vanilla 2D diffusion model as prior while being agnostic to the multi-view correlations, their results are sensitive to the initialization and hyperparameter settings. In contrast, our method directly adds 3D-aware control to the diffusion process, which essentially controls the supervisory of reconstruction’s 2D diffusion prior and consistently achieves high fidelity generation following users’ shape guidance. It is also noteworthy that, both Latent-NeRF and Fantasia3D require a long period of reconstruction (e.g., dozens of minutes) to give an impression of what the object might look like, making it unusable for interactive modeling, while our framework bypasses the reconstruction stage and allows to preview the 3D object in only a few seconds. 4.3 Interactive Generation with Part Editing We now present examples of interactive generation with progressive part editing. As shown in Fig. 6, users can first generate a basic instance (e.g., a base of cake, a teddy bear, or a penguin) with shape proxy, and then progressively add new shape blocks with changed text prompts (e.g., adding a small cake with candles, a green hat and red scarf, or even progressively add a torch and a red backpack from the back view), which seamlessly enrich the content of the instance while maintaining other parts unchanged. Notably, all these editing operations can be finished in roughly 5∼10 seconds, which then allows interactive previewing of the edited 3D results. Please refer to the supplementary video for the demonstration. 4.4 Ablation Studies Volume-conditioned reconstruction. We then inspect the efficacy of the volume-SDS loss by ablating during the shape reconstruction. As shown in Fig. 7, by adding volume-SDS loss, we achieve better geometry reconstruction (e.g., less floater and more reasonable chair bottom) than naïve training on fixed multiview images [Liu et al. 2023a; Long et al. 2023]. Proxy-bounded part editing. We finally analyze the proxy-bounded part editing by ablating proxy conditioning and mask dilation strategy in Fig. 8. Specifically, we choose a multi-step editing example, where the backpack should be edited from the back view. We merge the front and edited back image conditions using mixed denoising [Bar-Tal et al. 2023]. As shown in Fig. 8, the editing without the proxy condition would result in broken shapes (e.g., dangling flames and distorted body), while disabling mask dilation would also make the editing less natural (e.g., tightly fused red bag and broken hands). By equipping with the full strategies, we achieve a seamless editing effect while preserving other content unchanged. More ablation studies can be found in the supplementary material. Authors:
(1) Wenqi Dong, from Zhejiang University, and conducted this work during his internship at PICO, ByteDance;
(2) Bangbang Yang, from ByteDance contributed equally to this work together with Wenqi Dong;
(3) Lin Ma, ByteDance;
(4) Xiao Liu, ByteDance;
(5) Liyuan Cui, Zhejiang University;
(6) Hujun Bao, Zhejiang University;
(7) Yuewen Ma, ByteDance;
(8) Zhaopeng Cui, a Corresponding author from Zhejiang University. This paper is available on arxiv under CC BY 4.0 DEED license. Table of Links Abstract and 1 Introduction Abstract and 1 Introduction 2 Related Works 2 Related Works 3 Method and 3.1 Proxy-Guided 3D Conditioning for Diffusion 3 Method and 3.1 Proxy-Guided 3D Conditioning for Diffusion 3.2 Interactive Generation Workflow and 3.3 Volume Conditioned Reconstruction 3.2 Interactive Generation Workflow and 3.3 Volume Conditioned Reconstruction 4 Experiment and 4.1 Comparison on Proxy-based and Image-based 3D Generation 4 Experiment and 4.1 Comparison on Proxy-based and Image-based 3D Generation 4.2 Comparison on Controllable 3D Object Generation, 4.3 Interactive Generation with Part Editing & 4.4 Ablation Studies 4.2 Comparison on Controllable 3D Object Generation, 4.3 Interactive Generation with Part Editing & 4.4 Ablation Studies 5 Conclusions, Acknowledgments, and References 5 Conclusions, Acknowledgments, and References SUPPLEMENTARY MATERIAL SUPPLEMENTARY MATERIAL A. Implementation Details A. Implementation Details B. More Discussions B. More Discussions C. More Experiments C. More Experiments 4.2 Comparison on Controllable 3D Object Generation We compare our method with Controllable 3D object generation methods, including Latent-NeRF [Metzer et al. 2023] and Fantasia3D [Chen et al. 2023a]. Latent-NeRF introduces a sketch-shape guided loss, which constrains the density field close to the surface of the shape proxy, while Fantasia3D uses coarse shapes as the geometry initialization of DMTet [Shen et al. 2021]. In the experiment, we give all the methods the same coarse shape and the text prompts as a guidance and output the neural reconstructions’ rendered views for comparison (since Latent-NeRF does not officially provide mesh extraction). As shown in Fig. 4, Latent-NeRF generally obtains plausible results but fails to produce tiny shape and appearance details (e.g., blurry textured sofa, fox missing clear eyes and right arms and missing sunflower in Fig. 5 (b)), which indicates that directly applying 3D control to the 3D representation is sub-optimal since it might not work smoothly with the generic SDS loss. For Fantasia3D, since it only uses the 3D shape for initialization rather than supervision, it often generates overgrowth results that do not follow the given shape (e.g., inflate sofa, the fox with incomplete downward arms, pot with many leaves and a broken sunflower in Fig. 5 (c)) and also slightly suffers from “multi-face Janus problem” (e.g., multi-face fox in Fig. 5 (c)). Since both Latent-NeRF and Fantasia3D use vanilla 2D diffusion model as prior while being agnostic to the multi-view correlations, their results are sensitive to the initialization and hyperparameter settings. In contrast, our method directly adds 3D-aware control to the diffusion process, which essentially controls the supervisory of reconstruction’s 2D diffusion prior and consistently achieves high fidelity generation following users’ shape guidance. It is also noteworthy that, both Latent-NeRF and Fantasia3D require a long period of reconstruction (e.g., dozens of minutes) to give an impression of what the object might look like, making it unusable for interactive modeling, while our framework bypasses the reconstruction stage and allows to preview the 3D object in only a few seconds. 4.3 Interactive Generation with Part Editing We now present examples of interactive generation with progressive part editing. As shown in Fig. 6, users can first generate a basic instance (e.g., a base of cake, a teddy bear, or a penguin) with shape proxy, and then progressively add new shape blocks with changed text prompts (e.g., adding a small cake with candles, a green hat and red scarf, or even progressively add a torch and a red backpack from the back view), which seamlessly enrich the content of the instance while maintaining other parts unchanged. Notably, all these editing operations can be finished in roughly 5∼10 seconds, which then allows interactive previewing of the edited 3D results. Please refer to the supplementary video for the demonstration. 4.4 Ablation Studies Volume-conditioned reconstruction. We then inspect the efficacy of the volume-SDS loss by ablating during the shape reconstruction. As shown in Fig. 7, by adding volume-SDS loss, we achieve better geometry reconstruction (e.g., less floater and more reasonable chair bottom) than naïve training on fixed multiview images [Liu et al. 2023a; Long et al. 2023]. Proxy-bounded part editing. We finally analyze the proxy-bounded part editing by ablating proxy conditioning and mask dilation strategy in Fig. 8. Specifically, we choose a multi-step editing example, where the backpack should be edited from the back view. We merge the front and edited back image conditions using mixed denoising [Bar-Tal et al. 2023]. As shown in Fig. 8, the editing without the proxy condition would result in broken shapes (e.g., dangling flames and distorted body), while disabling mask dilation would also make the editing less natural (e.g., tightly fused red bag and broken hands). By equipping with the full strategies, we achieve a seamless editing effect while preserving other content unchanged. More ablation studies can be found in the supplementary material. Authors: (1) Wenqi Dong, from Zhejiang University, and conducted this work during his internship at PICO, ByteDance; (2) Bangbang Yang, from ByteDance contributed equally to this work together with Wenqi Dong; (3) Lin Ma, ByteDance; (4) Xiao Liu, ByteDance; (5) Liyuan Cui, Zhejiang University; (6) Hujun Bao, Zhejiang University; (7) Yuewen Ma, ByteDance; (8) Zhaopeng Cui, a Corresponding author from Zhejiang University. Authors: Authors: (1) Wenqi Dong, from Zhejiang University, and conducted this work during his internship at PICO, ByteDance; (2) Bangbang Yang, from ByteDance contributed equally to this work together with Wenqi Dong; (3) Lin Ma, ByteDance; (4) Xiao Liu, ByteDance; (5) Liyuan Cui, Zhejiang University; (6) Hujun Bao, Zhejiang University; (7) Yuewen Ma, ByteDance; (8) Zhaopeng Cui, a Corresponding author from Zhejiang University. This paper is available on arxiv under CC BY 4.0 DEED license. This paper is available on arxiv under CC BY 4.0 DEED license. available on arxiv available on arxiv

Coin3D Achieves Superior Control and Efficiency in 3D Generation

Regionalize APIs Like a Pro: Achieve Global Compliance and Scalability

Регионализирайте API като професионалист: Постигнете глобално съответствие и скалируемост

Regionalizirajte API-je kao profesionalac: Postignite globalnu usklađenost i skalabilnost

Περιφερειοποίηση APIs Like a Pro: Επίτευξη Παγκόσμιας Συμμόρφωσης και Επεκτασιμότητας

Regionalice las API como un profesional: logre cumplimiento y escalabilidad globales

プロのように API を地域化: グローバルなコンプライアンスとスケーラビリティを実現

Ampifandraisina amin'ny faritra toy ny Pro: Manatratra ny fanarahan-dalàna sy ny fahavitrihana maneran-tany

Регионизирајте ги API-ите како професионалец: Постигнете глобална усогласеност и приспособливост

Regionalizujte rozhrania API ako profesionál: Dosiahnite globálnu zhodu a škálovateľnosť

Регионализујте АПИ-је као професионалац: Постигните глобалну усклађеност и скалабилност

Минтақасозӣ кардани API-ҳо ба монанди Pro: Ба мувофиқати глобалӣ ва миқёспазирӣ ноил шавед

APIs کو ایک پرو کی طرح علاقائی بنائیں: عالمی تعمیل اور توسیع پذیری حاصل کریں۔

The Hydra Doom Tournament is a high-stakes gaming event, held both online and live, that highlights the classic game Doom while demonstrating the scalability and capabilities of the Hydra solution on the Cardano blockchain. The first stages of the tournament have delivered decisive results, with Cardano being crowned the fastest blockchain, achieving an unprecedented 1 million Transactions Per Second (TPS). This significant milestone positions Cardano at the forefront of blockchain scalability and could set a new precedent for the industry. Most importantly, Cardano has proven its capability for high scalability without compromising security or decentralization, reflecting the blockchain’s readiness to support high-demand applications across industries. Beyond the gaming industry, the Hydra Doom engine could transform P2P applications and business-to-business systems in all areas of tech. The Stress Test of the Hydra Doom Tournament ** ** The Hydra Doom Tournament has served as a real-world stress test that pushed Hydra’s capabilities to the limit within a competitive environment. The players competed in regional qualifiers and semi-finals, and the tournament will conclude with an in-person finale held in Las Vegas. The large scale of the tournament showcased Hydra’s ability to process vast amounts of data in real-time transactions. Superior Tech Applications in Data Processing Within the Hydra Doom Tournament, game data from players was submitted as smart contract transactions to individual Hydra heads. Each Hydra head then validated transactions using smart contracts that contained a subset of game logic and tracked the game state, including player scores. Providing this information via the blockchain made it possible to populate leaderboards and distribute prizes through verified smart contracts. Although video game competitions can typically take weeks to distribute prize money to winners, the Hydra Doom Tournament managed it within seconds. A Proving Ground for Scalability The Hydra Doom Tournament demonstrates Hydra’s scalability prowess under live conditions. During the qualifier on December 3rd, thousands of players competed for spots, so the Hydra team decided to further supplement traffic with additional games running bot-vs-bot simulations. The number of transactions surpassed the previous total of 106 million over 78 hours during the Rare Evo event, reaching 15.5 billion transactions in just 24 hours. The semi-finals, held December 8–18, featured Player-Vs-Player death matches as competitors vied for spots in the finals. A Major Milestone for Blockchain Technology ** ** Hydra’s success, proven by how rapidly it handled transactions throughout the Doom Tournament, has made Cardano much faster and more efficient. Hydra has distinguished itself from other blockchain systems through a number of advantages inherited from Cardano, including the EUTxO model, native assets, deterministic transaction execution, and high-assurance smart contracts. These benefits are taken to a new level with the application of the temporary head model. The Power of the Hydra Heads Hydra’s strength lies in its horizontal scaling capacity, in which as many Hydra heads as needed can be deployed in parallel. This grants Hydra the capability to take on large volumes, as proven on the Hydra Doom Tournament grounds. Theoretically, there is no upper limit to its scaling capacity for applications in this design type. The collaboration of multiple engineering teams and designers, including Sundae Labs, Adam Dean, DripDropz, and the Hydra team, has allowed for Hydra’s true potential to be unlocked and run in real time. What truly sets Hydra apart is the ingenuity of the temporary head model. Each Hydra head operator can enact trade-offs that wouldn’t be possible for persistent systems, such as setting transaction fees to zero, which allows for significantly larger and more complex transactions if needed. Perhaps most impressively, of the roughly 15.5 billion transactions processed within 24 hours on December 3, not a single one failed. This unprecedented reliability could set new benchmarks for the tech industry, with applications spanning multiple sectors. The Potential Applications of Cardano Hydra Hydra has already proven its potential in gaming through the Hydra Doom Tournament. Beyond processing transactions for these kinds of tournaments, it could also process microtransactions during gameplay with advanced speed and efficiency. These capabilities can be taken outside the gaming world as well, as it is uniquely suited for P2P applications within business-to-business systems. Broadening the category from the gaming angle, it can also be applied to “information gathering” applications, where the primary purpose is not to handle people’s assets but to collect cryptographically verified information such as signed messages. Furthermore, Hydra’s abilities could be utilized for banking and global payments, breaking financial barriers with faster cross-border transactions. Within the world of DeFi, Hydra could power decentralized applications to improve the user experience among a wide variety of other blockchain use cases. The Future Implications of Blockchain Progression Each milestone of blockchain progression increases the accessibility of blockchain technology for users and use cases. As this technology reaches new levels of sophistication, so does its potential to transform industries across tech. In terms of Hydra, it can be employed as a “friction remover” for practical day-to-day use. In its initial deployment phase, it can operate in a centralized model, then gradually change its structure to involve more node operators. Although Hydra is not the ultimate solution for every tech case, it can be combined with other products in the ecosystem to boost value for businesses and organizations. The Road Ahead for Cardano Hydra The next step for Hydra’s team is the conclusion of the Hydra Doom Tournament, which will take place on January 24, 2025 during the C-PLAY SUMMIT at the Hyper X Arena in Las Vegas. In a final showdown, top players will compete for a total of $50,000 in prizes. This showcase event will place Hydra’s scalability on the global stage in true esports fashion. Beyond the tournament, the Hydra team looks forward to the system’s official version 1 node launch. They also have numerous applications in the works that they will build with Hydra, including notable applications aimed at empowering DeFi use cases with Gummiworm from Sundae Labs and Delta DeFi. With the capabilities that Cardano Hydra has demonstrated so far, it is poised to take top spots in blockchain applications. Like its namesake, the multi-headed mythical beast, Hydra’s decentralized mode of operation makes it an exciting example of what DeFi has to offer. The Hydra Doom Tournament is a high-stakes gaming event, held both online and live, that highlights the classic game Doom while demonstrating the scalability and capabilities of the Hydra solution on the Cardano blockchain. The first stages of the tournament have delivered decisive results, with Cardano being crowned the fastest blockchain, achieving an unprecedented 1 million Transactions Per Second (TPS). Cardano Cardano This significant milestone positions Cardano at the forefront of blockchain scalability and could set a new precedent for the industry. Most importantly, Cardano has proven its capability for high scalability without compromising security or decentralization, reflecting the blockchain’s readiness to support high-demand applications across industries. Beyond the gaming industry, the Hydra Doom engine could transform P2P applications and business-to-business systems in all areas of tech. The Stress Test of the Hydra Doom Tournament The Stress Test of the Hydra Doom Tournament ** ** The Hydra Doom Tournament has served as a real-world stress test that pushed Hydra’s capabilities to the limit within a competitive environment. The players competed in regional qualifiers and semi-finals, and the tournament will conclude with an in-person finale held in Las Vegas. The large scale of the tournament showcased Hydra’s ability to process vast amounts of data in real-time transactions. Superior Tech Applications in Data Processing Superior Tech Applications in Data Processing Within the Hydra Doom Tournament, game data from players was submitted as smart contract transactions to individual Hydra heads. Each Hydra head then validated transactions using smart contracts that contained a subset of game logic and tracked the game state, including player scores. Providing this information via the blockchain made it possible to populate leaderboards and distribute prizes through verified smart contracts. Although video game competitions can typically take weeks to distribute prize money to winners, the Hydra Doom Tournament managed it within seconds. A Proving Ground for Scalability A Proving Ground for Scalability The Hydra Doom Tournament demonstrates Hydra’s scalability prowess under live conditions. During the qualifier on December 3rd, thousands of players competed for spots, so the Hydra team decided to further supplement traffic with additional games running bot-vs-bot simulations. The number of transactions surpassed the previous total of 106 million over 78 hours during the Rare Evo event, reaching 15.5 billion transactions in just 24 hours. The semi-finals, held December 8–18, featured Player-Vs-Player death matches as competitors vied for spots in the finals. A Major Milestone for Blockchain Technology ** ** Hydra’s success, proven by how rapidly it handled transactions throughout the Doom Tournament, has made Cardano much faster and more efficient. Hydra has distinguished itself from other blockchain systems through a number of advantages inherited from Cardano, including the EUTxO model, native assets, deterministic transaction execution, and high-assurance smart contracts. These benefits are taken to a new level with the application of the temporary head model. The Power of the Hydra Heads The Power of the Hydra Heads Hydra’s strength lies in its horizontal scaling capacity, in which as many Hydra heads as needed can be deployed in parallel. This grants Hydra the capability to take on large volumes, as proven on the Hydra Doom Tournament grounds. Theoretically, there is no upper limit to its scaling capacity for applications in this design type. The collaboration of multiple engineering teams and designers, including Sundae Labs, Adam Dean, DripDropz, and the Hydra team, has allowed for Hydra’s true potential to be unlocked and run in real time. What truly sets Hydra apart is the ingenuity of the temporary head model. Each Hydra head operator can enact trade-offs that wouldn’t be possible for persistent systems, such as setting transaction fees to zero, which allows for significantly larger and more complex transactions if needed. Perhaps most impressively, of the roughly 15.5 billion transactions processed within 24 hours on December 3, not a single one failed. This unprecedented reliability could set new benchmarks for the tech industry, with applications spanning multiple sectors. The Potential Applications of Cardano Hydra The Potential Applications of Cardano Hydra Hydra has already proven its potential in gaming through the Hydra Doom Tournament. Beyond processing transactions for these kinds of tournaments, it could also process microtransactions during gameplay with advanced speed and efficiency. These capabilities can be taken outside the gaming world as well, as it is uniquely suited for P2P applications within business-to-business systems. Broadening the category from the gaming angle, it can also be applied to “information gathering” applications, where the primary purpose is not to handle people’s assets but to collect cryptographically verified information such as signed messages. Furthermore, Hydra’s abilities could be utilized for banking and global payments, breaking financial barriers with faster cross-border transactions. Within the world of DeFi, Hydra could power decentralized applications to improve the user experience among a wide variety of other blockchain use cases. The Future Implications of Blockchain Progression The Future Implications of Blockchain Progression Each milestone of blockchain progression increases the accessibility of blockchain technology for users and use cases. As this technology reaches new levels of sophistication, so does its potential to transform industries across tech. In terms of Hydra, it can be employed as a “friction remover” for practical day-to-day use. In its initial deployment phase, it can operate in a centralized model, then gradually change its structure to involve more node operators. Although Hydra is not the ultimate solution for every tech case, it can be combined with other products in the ecosystem to boost value for businesses and organizations. The Road Ahead for Cardano Hydra The Road Ahead for Cardano Hydra The next step for Hydra’s team is the conclusion of the Hydra Doom Tournament, which will take place on January 24, 2025 during the C-PLAY SUMMIT at the Hyper X Arena in Las Vegas. In a final showdown, top players will compete for a total of $50,000 in prizes. This showcase event will place Hydra’s scalability on the global stage in true esports fashion. C-PLAY SUMMIT C-PLAY SUMMIT Beyond the tournament, the Hydra team looks forward to the system’s official version 1 node launch. They also have numerous applications in the works that they will build with Hydra, including notable applications aimed at empowering DeFi use cases with Gummiworm from Sundae Labs and Delta DeFi. Hydra Hydra With the capabilities that Cardano Hydra has demonstrated so far, it is poised to take top spots in blockchain applications. Like its namesake, the multi-headed mythical beast, Hydra’s decentralized mode of operation makes it an exciting example of what DeFi has to offer.

Hydra Achieves 1 Million TPS, Validating Cardano as the Fastest Blockchain

Hydra aconsegueix 1 milió de TPS, validant Cardano com la cadena de blocs més ràpida

Hydra alcanza 1 millón de TPS, lo que valida a Cardano como la cadena de bloques más rápida

Hydra saavuttaa 1 miljoonan TPS:n, vahvistaen Cardanon nopeimmaksi lohkoketjuksi

हाइड्रा ने 1 मिलियन टीपीएस हासिल किया, कार्डानो को सबसे तेज ब्लॉकचेन के रूप में मान्यता दी

Hydra postiže 1 milijun TPS-a, potvrđujući Cardano kao najbrži blockchain

Hydra reyalize 1 milyon TPS, valide Cardano kòm Blockchain ki pi rapid la

Hydraが100万TPSを達成、Cardanoが最速ブロックチェーンであることを実証

Nahazo TPS 1 tapitrisa i Hydra, manamarina an'i Cardano ho Blockchain haingana indrindra

Hydra osiąga 1 milion TPS, potwierdzając, że Cardano jest najszybszym blockchainem

Hydra atinge 1 milhão de TPS, validando Cardano como o blockchain mais rápido

Hydra досягає 1 мільйона TPS, підтверджуючи Cardano як найшвидший блокчейн

Hydra 实现 100 万 TPS，证实 Cardano 是速度最快的区块链

Author:
(1) Subhajit Sahu, IIIT Hyderabad, Hyderabad, Telangana, India (subhajit.sahu@research.iiit.ac.in). Table of Links Abstract and 1 Introduction 2 Related Work 3 Preliminaries 4 Approach 5.1 Experimental Setup 5.2 Performance of Dynamic Frontier PageRank 5.3 Strong Scaling of Dynamic Frontier PageRan 6 Conclusion, Acknowledgments, and References 5. EVALUATION 5.1 Experimental Setup 5.1.1 System used. We conduct experiments on a system equipped with an AMD EPYC-7742 processor, with 64 cores and operating at a frequency of 2.25 GHz. Each core has a 4 MB L1 cache, a 32 MB L2 cache, and shares a 256 MB L3 cache. The server is configured with 512 GB of DDR4 system memory and operates on Ubuntu 20.04. 5.1.3 Dataset. We use four graph classes sourced from the SuiteSparse Matrix Collection [14], as detailed in Table 1. The number of vertices in these graphs range from 3.07 million to 214 million, with edge counts spanning from 37.4 million to 1.98 billion. To address the impact of dead ends (vertices lacking out-links), a global teleport rank computation is needed in each iteration. We mitigate this overhead by adding self-loops to all vertices in the graph [1, 15]. This paper is available on arxiv under CC BY-NC-SA 4.0 DEED license. Author: (1) Subhajit Sahu, IIIT Hyderabad, Hyderabad, Telangana, India (subhajit.sahu@research.iiit.ac.in). Author: Author: (1) Subhajit Sahu, IIIT Hyderabad, Hyderabad, Telangana, India (subhajit.sahu@research.iiit.ac.in). Table of Links Abstract and 1 Introduction Abstract and 1 Introduction 2 Related Work 2 Related Work 3 Preliminaries 3 Preliminaries 4 Approach 4 Approach 5.1 Experimental Setup 5.1 Experimental Setup 5.2 Performance of Dynamic Frontier PageRank 5.2 Performance of Dynamic Frontier PageRank 5.3 Strong Scaling of Dynamic Frontier PageRan 5.3 Strong Scaling of Dynamic Frontier PageRan 6 Conclusion, Acknowledgments, and References 6 Conclusion, Acknowledgments, and References 5. EVALUATION 5.1 Experimental Setup 5.1.1 System used. We conduct experiments on a system equipped with an AMD EPYC-7742 processor, with 64 cores and operating at a frequency of 2.25 GHz. Each core has a 4 MB L1 cache, a 32 MB L2 cache, and shares a 256 MB L3 cache. The server is configured with 512 GB of DDR4 system memory and operates on Ubuntu 20.04. 5.1.1 System used. 5.1.3 Dataset. We use four graph classes sourced from the SuiteSparse Matrix Collection [14], as detailed in Table 1. The number of vertices in these graphs range from 3.07 million to 214 million, with edge counts spanning from 37.4 million to 1.98 billion. To address the impact of dead ends (vertices lacking out-links), a global teleport rank computation is needed in each iteration. We mitigate this overhead by adding self-loops to all vertices in the graph [1, 15]. 5.1.3 Dataset. This paper is available on arxiv under CC BY-NC-SA 4.0 DEED license. This paper is available on arxiv under CC BY-NC-SA 4.0 DEED license. available on arxiv

Dynamic Frontier PageRank Achieves Efficiency and Accuracy with Batch Updates and Parallel Computing

Author:
(1) Subhajit Sahu, IIIT Hyderabad, Hyderabad, Telangana, India (subhajit.sahu@research.iiit.ac.in). Table of Links Abstract and 1 Introduction 2 Related Work 3 Preliminaries 4 Approach 5.1 Experimental Setup 5.2 Performance of Dynamic Frontier PageRank 5.3 Strong Scaling of Dynamic Frontier PageRan 6 Conclusion, Acknowledgments, and References 5.3 Strong Scaling of Dynamic Frontier PageRan This paper is available on arxiv under CC BY-NC-SA 4.0 DEED license. Author: (1) Subhajit Sahu, IIIT Hyderabad, Hyderabad, Telangana, India (subhajit.sahu@research.iiit.ac.in). Author: Author: (1) Subhajit Sahu, IIIT Hyderabad, Hyderabad, Telangana, India (subhajit.sahu@research.iiit.ac.in). Table of Links Abstract and 1 Introduction Abstract and 1 Introduction 2 Related Work 2 Related Work 3 Preliminaries 3 Preliminaries 4 Approach 4 Approach 5.1 Experimental Setup 5.1 Experimental Setup 5.2 Performance of Dynamic Frontier PageRank 5.2 Performance of Dynamic Frontier PageRank 5.3 Strong Scaling of Dynamic Frontier PageRan 5.3 Strong Scaling of Dynamic Frontier PageRan 6 Conclusion, Acknowledgments, and References 6 Conclusion, Acknowledgments, and References 5.3 Strong Scaling of Dynamic Frontier PageRan This paper is available on arxiv under CC BY-NC-SA 4.0 DEED license. This paper is available on arxiv under CC BY-NC-SA 4.0 DEED license. available on arxiv

Strong Scaling Achieves 15.2× Speedup for Dynamic Graph Updates with Multi-Threaded Efficiency

When I Implemented batch generation of DolphinScheduler tasks and imported them, it was found that tasks could only be imported one by one, so using the API is apparently more convenient. DolphinScheduler API Documentation DolphinScheduler has API documentation available at: http://IP:12345/dolphinscheduler/swagger-ui/index.html?language=zh_CN&lang=cn However, the documentation is relatively brief, so you need to study it yourself. Token: All APIs require a token. In Security Center - Token Management, create a token. Remember this token, as it will be needed for all subsequent API calls. Header: Construct the request header using the token mentioned above: token = ''
headers = {
    'Accept': 'application/json',
    'token': token
} Project ID: project_id can be found in the URL when viewing the project workflow. DolphinScheduler Task Import API The API for importing tasks is: import_url = 'http://IP:12345/dolphinscheduler/projects/{project_id}/process-definition/import' Knowing the API, you can proceed with the import: def import_job(file_path):
# Open the file and read it as binary data
    with open(file_path, 'rb') as file:
        files = {'file': file}
        # Import the workflow
        response = requests.post(import_url, headers=headers, files=files)
        print(response.status_code)
        if response.status_code != 200:
            print('Upload failed  ' + file_path) It should be noted that task imports only support binary files. file_path is the workflow file. For implementation, you can export one from the workflow as a reference.By repeating the above method, you can achieve batch task imports. Workflow Deployment After completing batch task uploads using the above method, there is still an issue: manually deploying each workflow is still a significant workload. Therefore, we continue using the API. After research, it was found that deploying a workflow requires first obtaining the schedule ID of the workflow. Get Workflow List -> Get Workflow Code -> Get Schedule IDs for All Workflows -> Deploy Workflows Get Workflow ListThe API address is: jobs_url = 'http://IP:12345/dolphinscheduler/projects/{project_id}/process-definition' However, this requires paginated queries, which is slightly inconvenient: def get_jobs_list():
    # Paginated query
    # Initialize pagination parameters
    pageNo = 1
    pageSize = 10
    url = f'{jobs_url}?pageSize=10&pageNo=1&searchVal='
    # Construct the complete URL
    # Store all results
    all_items = list()
    while True:
        # Construct the complete URL
        url = f'{jobs_url}?pageSize={pageSize}&pageNo={pageNo}&searchVal='

        # Send GET request
        response = requests.get(url, headers=headers)

        # Check response status code
        if response.status_code == 200:
            # Request succeeded, process response data
            items = response.content.decode()
            total = json.loads(items)["data"]["total"]
            item = json.loads(items)["data"]["totalList"]
            # Add the current page's data to the result list
            for i in item:
                all_items.append(i)

            # Exit the loop if there is no data on the current page
            if pageNo * pageSize > total:
                break
            if not items:
                break
            # Increment page number
            pageNo += 1
        else:
            # Request failed, print error information
            print('Request failed:', response.status_code, response.text)
            break

    return all_items all_items contains the specific details of all workflows, which need to be extracted: all_jobs = get_jobs_list()
        job_codes = [job['code'] for job in all_jobs] This gives you the codes for all workflows. Get Schedule ID：The API for schedule IDs is as follows. To avoid pagination, 1000 entries per page are directly specified: schedules_url = 'http://36.133.140.132:12345/dolphinscheduler/projects/{project_id}/schedules?pageSize=1000&pageNo=1&processDefinitionCode=' Using this API, you can obtain all schedule IDs: def schedule_id(job_code):
    url = schedules_url + str(job_code)
    response = requests.get(url, headers=headers)
    if response.status_code == 200:
        data = response.content.decode()
        js = json.loads(data)
        if len(js['data']['totalList']) > 0 and js['data']['totalList'][0]['releaseState'] == 'OFFLINE':
            return js['data']['totalList'][0]['id']
    else:
        return '' Here, schedule IDs that are already deployed are filtered out. Deploy：Everything is ready, and deployment can finally proceed: online_url = 'http://36.133.140.132:12345/dolphinscheduler/projects/{project_id}/schedules/{scheduler_id}/online' Specific implementation: def online_job(scheduler_id):
    url = online_url.format(scheduler_id=scheduler_id)
    response = requests.post(url, headers=headers)
    if response.status_code == 200:
        print('success')
    else:
        print('online job failed') At this point, import and fully automated batch deployment can be achieved. When I Implemented batch generation of DolphinScheduler tasks and imported them, it was found that tasks could only be imported one by one, so using the API is apparently more convenient. DolphinScheduler API Documentation DolphinScheduler has API documentation available at: http://IP:12345/dolphinscheduler/swagger-ui/index.html?language=zh_CN&lang=cn http://IP:12345/dolphinscheduler/swagger-ui/index.html?language=zh_CN&lang=cn However, the documentation is relatively brief, so you need to study it yourself. Token: All APIs require a token. Token: In Security Center - Token Management , create a token. Remember this token, as it will be needed for all subsequent API calls. Security Center - Token Management Header: Construct the request header using the token mentioned above: Header: token = ''
headers = {
    'Accept': 'application/json',
    'token': token
} token = ''
headers = {
    'Accept': 'application/json',
    'token': token
} Project ID: project_id can be found in the URL when viewing the project workflow. Project ID: project_id DolphinScheduler Task Import API The API for importing tasks is: import_url = 'http://IP:12345/dolphinscheduler/projects/{project_id}/process-definition/import' import_url = 'http://IP:12345/dolphinscheduler/projects/{project_id}/process-definition/import' Knowing the API, you can proceed with the import: def import_job(file_path):
# Open the file and read it as binary data
    with open(file_path, 'rb') as file:
        files = {'file': file}
        # Import the workflow
        response = requests.post(import_url, headers=headers, files=files)
        print(response.status_code)
        if response.status_code != 200:
            print('Upload failed  ' + file_path) def import_job(file_path):
# Open the file and read it as binary data
    with open(file_path, 'rb') as file:
        files = {'file': file}
        # Import the workflow
        response = requests.post(import_url, headers=headers, files=files)
        print(response.status_code)
        if response.status_code != 200:
            print('Upload failed  ' + file_path) It should be noted that task imports only support binary files. file_path is the workflow file. For implementation, you can export one from the workflow as a reference.By repeating the above method, you can achieve batch task imports. file_path Workflow Deployment After completing batch task uploads using the above method, there is still an issue: manually deploying each workflow is still a significant workload. Therefore, we continue using the API. After research, it was found that deploying a workflow requires first obtaining the schedule ID of the workflow. Get Workflow List -> Get Workflow Code -> Get Schedule IDs for All Workflows -> Deploy Workflows Get Workflow List -> Get Workflow Code -> Get Schedule IDs for All Workflows -> Deploy Workflows Get Workflow List -> Get Workflow Code -> Get Schedule IDs for All Workflows -> Deploy Workflows Get Workflow ListThe API address is: Get Workflow List The API address is: Get Workflow List jobs_url = 'http://IP:12345/dolphinscheduler/projects/{project_id}/process-definition' jobs_url = 'http://IP:12345/dolphinscheduler/projects/{project_id}/process-definition' However, this requires paginated queries, which is slightly inconvenient: def get_jobs_list():
    # Paginated query
    # Initialize pagination parameters
    pageNo = 1
    pageSize = 10
    url = f'{jobs_url}?pageSize=10&pageNo=1&searchVal='
    # Construct the complete URL
    # Store all results
    all_items = list()
    while True:
        # Construct the complete URL
        url = f'{jobs_url}?pageSize={pageSize}&pageNo={pageNo}&searchVal='

        # Send GET request
        response = requests.get(url, headers=headers)

        # Check response status code
        if response.status_code == 200:
            # Request succeeded, process response data
            items = response.content.decode()
            total = json.loads(items)["data"]["total"]
            item = json.loads(items)["data"]["totalList"]
            # Add the current page's data to the result list
            for i in item:
                all_items.append(i)

            # Exit the loop if there is no data on the current page
            if pageNo * pageSize > total:
                break
            if not items:
                break
            # Increment page number
            pageNo += 1
        else:
            # Request failed, print error information
            print('Request failed:', response.status_code, response.text)
            break

    return all_items def get_jobs_list():
    # Paginated query
    # Initialize pagination parameters
    pageNo = 1
    pageSize = 10
    url = f'{jobs_url}?pageSize=10&pageNo=1&searchVal='
    # Construct the complete URL
    # Store all results
    all_items = list()
    while True:
        # Construct the complete URL
        url = f'{jobs_url}?pageSize={pageSize}&pageNo={pageNo}&searchVal='

        # Send GET request
        response = requests.get(url, headers=headers)

        # Check response status code
        if response.status_code == 200:
            # Request succeeded, process response data
            items = response.content.decode()
            total = json.loads(items)["data"]["total"]
            item = json.loads(items)["data"]["totalList"]
            # Add the current page's data to the result list
            for i in item:
                all_items.append(i)

            # Exit the loop if there is no data on the current page
            if pageNo * pageSize > total:
                break
            if not items:
                break
            # Increment page number
            pageNo += 1
        else:
            # Request failed, print error information
            print('Request failed:', response.status_code, response.text)
            break

    return all_items all_items contains the specific details of all workflows, which need to be extracted: all_items all_jobs = get_jobs_list()
        job_codes = [job['code'] for job in all_jobs] all_jobs = get_jobs_list()
        job_codes = [job['code'] for job in all_jobs] This gives you the codes for all workflows. Get Schedule ID：The API for schedule IDs is as follows. To avoid pagination, 1000 entries per page are directly specified: Get Schedule ID：The API for schedule IDs is as follows. To avoid pagination, 1000 entries per page are directly specified: schedules_url = 'http://36.133.140.132:12345/dolphinscheduler/projects/{project_id}/schedules?pageSize=1000&pageNo=1&processDefinitionCode=' schedules_url = 'http://36.133.140.132:12345/dolphinscheduler/projects/{project_id}/schedules?pageSize=1000&pageNo=1&processDefinitionCode=' Using this API, you can obtain all schedule IDs: def schedule_id(job_code):
    url = schedules_url + str(job_code)
    response = requests.get(url, headers=headers)
    if response.status_code == 200:
        data = response.content.decode()
        js = json.loads(data)
        if len(js['data']['totalList']) > 0 and js['data']['totalList'][0]['releaseState'] == 'OFFLINE':
            return js['data']['totalList'][0]['id']
    else:
        return '' def schedule_id(job_code):
    url = schedules_url + str(job_code)
    response = requests.get(url, headers=headers)
    if response.status_code == 200:
        data = response.content.decode()
        js = json.loads(data)
        if len(js['data']['totalList']) > 0 and js['data']['totalList'][0]['releaseState'] == 'OFFLINE':
            return js['data']['totalList'][0]['id']
    else:
        return '' Here, schedule IDs that are already deployed are filtered out. Deploy：Everything is ready, and deployment can finally proceed: Deploy：Everything is ready, and deployment can finally proceed: online_url = 'http://36.133.140.132:12345/dolphinscheduler/projects/{project_id}/schedules/{scheduler_id}/online' online_url = 'http://36.133.140.132:12345/dolphinscheduler/projects/{project_id}/schedules/{scheduler_id}/online' Specific implementation: def online_job(scheduler_id):
    url = online_url.format(scheduler_id=scheduler_id)
    response = requests.post(url, headers=headers)
    if response.status_code == 200:
        print('success')
    else:
        print('online job failed') def online_job(scheduler_id):
    url = online_url.format(scheduler_id=scheduler_id)
    response = requests.post(url, headers=headers)
    if response.status_code == 200:
        print('success')
    else:
        print('online job failed') At this point, import and fully automated batch deployment can be achieved.

Using DolphinScheduler API to Achieve Efficient Batch Workflow Import and Script Deployment

How Apex Fusion Achieved FINMA Qualification Without Compromising Decentralization

According to the Institute of International Education, the United States hosted more than 1.1 million international students in the 2023-2024 academic year. Although this number has increased yearly since the COVID-19 pandemic, these latest numbers mark an all-time high. That being said, even though the United States has become a leading destination for international students, especially in the technology sector, students who wish to journey to the land of red, white, and blue must go through a long and complicated process before they can enroll in the school of their dreams. While many people discuss the challenges of international studies, like language barriers and cultural differences, there are often even bigger challenges, such as applying for visas or finding a place to live during and after studies. Navigating the challenges of being an international student Thankfully, resources are available for students struggling with these challenges, and Otips provides a centralized resource for international students to simplify their educational and immigration journeys. Otips has centralized and streamlined the study abroad process for thousands of students globally, reducing administrative burdens and improving accessibility to U.S. educational opportunities. From enrollment to visa guidance, Otips offers personalized support that helps them navigate the study abroad process comfortably and easily. Otips was inspired by the experiences of its founder and CEO, Botir Nasridinov. “When I first arrived in the U.S. to attend a language school, I ended up overpaying for my education because I wasn’t familiar with the local system and available options,” he explains. “This lack of information inspired me to create a platform to help students find suitable schools. Recognizing a critical gap in resources for international students, Botir Nasridinov developed Otips, a first-of-its-kind platform that revolutionizes how students navigate complex processes like admissions, visa applications, and housing, fundamentally transforming the study abroad experience. “The challenges I faced as an international student — high living costs, difficulty selecting the right academic programs, and housing problems — motivated me to found ClassTrip,” he says. “My goal was to create a platform that provides students with access to affordable education, housing solutions, and healthcare advice, making studying abroad more accessible and less costly.” Under Botir Nasridinov’s leadership, Otips has been recognized as a groundbreaking platform in international education, forming strategic partnerships with over 200 U.S. educational institutions. “My mission is to remove barriers for students and help them achieve their dreams confidently and easily,” he says. How Otips helps students in their study abroad process This platform, ClassTrip, would soon evolve into the service known as Otips today. The primary service of Otips is its “education marketplace,” which helps students from around the world with counseling and admissions for over 200 educational institutions in multiple states in the United States, including language schools, colleges, and universities. This can help streamline the application process, which can be quite complex. Furthermore, Otips partners with several institutions to offer scholarships and tuition discounts of up to 70%, allowing students to achieve their dreams of studying internationally much more affordably. Beyond helping international students find and apply to the right schools, Otips can also assist with their visa applications. The team at Otips is experienced in a wide range of immigration cases, allowing them to ensure that applications are prepared accurately to maximize an applicant’s chance of approval. Applicants can also get support and guidance for other essential steps in the visa application process, including the all-important and often stressed-about interview. The other aspect of international study that can be difficult for students is finding housing. Many schools in the United States do not offer housing assistance to international students, leaving them on their own to navigate the process of finding a place to live in a country that they often have never been to before. It can sometimes take weeks or even months for international students to find safe, affordable housing situations. Otips offers resources that help students looking for accommodations connect with hosts, giving them greater value and allowing them to focus more on discovering their destination and enjoying their trip. Studying abroad is an incredibly enriching experience, especially for those who want to take advantage of America’s educational opportunities in fields where the country excels, such as technology. Navigating the application and visa process can be difficult, but with the help of services like Otips, students can get the support they need to have the study abroad experience of their dreams. According to the Institute of International Education , the United States hosted more than 1.1 million international students in the 2023-2024 academic year. Although this number has increased yearly since the COVID-19 pandemic, these latest numbers mark an all-time high. Institute of International Education Institute of International Education That being said, even though the United States has become a leading destination for international students, especially in the technology sector , students who wish to journey to the land of red, white, and blue must go through a long and complicated process before they can enroll in the school of their dreams. While many people discuss the challenges of international studies, like language barriers and cultural differences, there are often even bigger challenges, such as applying for visas or finding a place to live during and after studies. especially in the technology sector especially in the technology sector Navigating the challenges of being an international student Thankfully, resources are available for students struggling with these challenges, and Otips provides a centralized resource for international students to simplify their educational and immigration journeys. Otips has centralized and streamlined the study abroad process for thousands of students globally, reducing administrative burdens and improving accessibility to U.S. educational opportunities. From enrollment to visa guidance, Otips offers personalized support that helps them navigate the study abroad process comfortably and easily. Otips Otips Otips was inspired by the experiences of its founder and CEO, Botir Nasridinov. “When I first arrived in the U.S. to attend a language school, I ended up overpaying for my education because I wasn’t familiar with the local system and available options,” he explains. “This lack of information inspired me to create a platform to help students find suitable schools. Recognizing a critical gap in resources for international students, Botir Nasridinov developed Otips, a first-of-its-kind platform that revolutionizes how students navigate complex processes like admissions, visa applications, and housing, fundamentally transforming the study abroad experience. “The challenges I faced as an international student — high living costs, difficulty selecting the right academic programs, and housing problems — motivated me to found ClassTrip,” he says. “My goal was to create a platform that provides students with access to affordable education, housing solutions, and healthcare advice, making studying abroad more accessible and less costly.” Under Botir Nasridinov’s leadership, Otips has been recognized as a groundbreaking platform in international education, forming strategic partnerships with over 200 U.S. educational institutions. “My mission is to remove barriers for students and help them achieve their dreams confidently and easily,” he says. How Otips helps students in their study abroad process This platform, ClassTrip, would soon evolve into the service known as Otips today. The primary service of Otips is its “education marketplace,” which helps students from around the world with counseling and admissions for over 200 educational institutions in multiple states in the United States, including language schools, colleges, and universities. This can help streamline the application process, which can be quite complex. Furthermore, Otips partners with several institutions to offer scholarships and tuition discounts of up to 70%, allowing students to achieve their dreams of studying internationally much more affordably. Beyond helping international students find and apply to the right schools, Otips can also assist with their visa applications. The team at Otips is experienced in a wide range of immigration cases, allowing them to ensure that applications are prepared accurately to maximize an applicant’s chance of approval. Applicants can also get support and guidance for other essential steps in the visa application process, including the all-important and often stressed-about interview. The other aspect of international study that can be difficult for students is finding housing. Many schools in the United States do not offer housing assistance to international students, leaving them on their own to navigate the process of finding a place to live in a country that they often have never been to before. It can sometimes take weeks or even months for international students to find safe, affordable housing situations. Otips offers resources that help students looking for accommodations connect with hosts, giving them greater value and allowing them to focus more on discovering their destination and enjoying their trip. Studying abroad is an incredibly enriching experience, especially for those who want to take advantage of America’s educational opportunities in fields where the country excels, such as technology. Navigating the application and visa process can be difficult, but with the help of services like Otips, students can get the support they need to have the study abroad experience of their dreams.

Helping International Students Achieve Their Dream Study Abroad Experience

مساعدة الطلاب الدوليين على تحقيق أحلامهم في الدراسة بالخارج

Подпомагане на чуждестранни студенти да постигнат мечтаното обучение в чужбина

Ayudando a los estudiantes internacionales a lograr su sueño de estudiar en el extranjero

Axudando aos estudantes internacionais a lograr a súa experiencia de estudo no estranxeiro soñada

Pomaganje međunarodnim studentima da ostvare svoje iskustvo studiranja u inozemstvu iz snova

Օգնում է միջազգային ուսանողներին հասնել իրենց երազանքի ուսման արտերկրում

Membantu Pelajar Internasional Meraih Impian Mereka untuk Belajar di Luar Negeri

留学生の夢の留学体験の実現を支援

Ajudando estudantes internacionais a realizar a experiência dos sonhos de estudar no exterior

Помощь иностранным студентам в достижении их мечты об обучении за рубежом

Pomáhame zahraničným študentom dosiahnuť vysnívanú skúsenosť so štúdiom v zahraničí

Помагање међународним студентима да остваре своје искуство студирања у иностранству

Workflow Optimization in Healthcare Insurance: How We Achieved 88% Reduction in Execution Time

በጤና እንክብካቤ ኢንሹራንስ ውስጥ የስራ ፍሰት ማመቻቸት፡ በአፈፃፀም ጊዜ 88% ቅነሳን እንዴት እንዳሳካን።

Оптимизиране на работния процес в здравното осигуряване: Как постигнахме 88% намаление на времето за изпълнение

Optimizacija toka rada u zdravstvenom osiguranju: kako smo postigli 88% smanjenje vremena izvršenja

Optimering af arbejdsgange i sundhedsforsikring: Sådan opnåede vi 88 % reduktion i eksekveringstid

Optimización del flujo de trabajo en seguros médicos: cómo logramos una reducción del 88 % en el tiempo de ejecución

医療保険におけるワークフローの最適化: 実行時間を 88% 削減した方法

Otimização do fluxo de trabalho em seguros de saúde: como alcançamos uma redução de 88% no tempo de execução

Оптимизация рабочего процесса в медицинском страховании: как нам удалось сократить время выполнения на 88%

Optimalizácia pracovného toku v zdravotnom poistení: Ako sme dosiahli 88 % skrátenie času realizácie

Arbetsflödesoptimering i sjukvårdsförsäkringar: Hur vi uppnådde 88 % minskning av exekveringstid

Saglygy goraýyş ätiýaçlandyryşynda iş akymynyň optimizasiýasy: ecutionerine ýetiriş wagtynda 88% azalmagy nädip gazandyk

医疗保险中的工作流程优化：我们如何将执行时间缩短 88%

The HackerNoon Newsletter: Heres Why High Achievers Feel Like Failures (12/29/2024)

Table of Links Abstract and 1. Introduction 2. Preliminaries and 2.1. Blind deconvolution 2.2. Quadratic neural networks 3. Methodology 3.1. Time domain quadratic convolutional filter 3.2. Superiority of cyclic features extraction by QCNN 3.3. Frequency domain linear filter with envelope spectrum objective function 3.4. Integral optimization with uncertainty-aware weighing scheme 4. Computational experiments 4.1. Experimental configurations 4.2. Case study 1: PU dataset 4.3. Case study 2: JNU dataset 4.4. Case study 3: HIT dataset 5. Computational experiments 5.1. Comparison of BD methods 5.2. Classification results on various noise conditions 5.3. Employing ClassBD to deep learning classifiers 5.4. Employing ClassBD to machine learning classifiers 5.5. Feature extraction ability of quadratic and conventional networks 5.6. Comparison of ClassBD filters 6. Conclusions Appendix and References 4.4.1. Dataset description 4.4.1. Dataset description The Harbin Institute of Technology (HIT) dataset was collected by our team. The data acquisition for faulty bearings was conducted at the MIIT Key Laboratory of Aerospace Bearing Technology and Equipment at HIT. The bearing test rig and faulty bearings are illustrated in Figure 4. This test rig follows the Chinese standard rolling bearing measuring method (GB/T 32333-2015). The test bearings used were HC7003 angular contact ball bearings, typically employed in high-speed rotating machines. For signal collection, an acceleration sensor was directly attached to the bearing, and the NI USB-6002 device was used at a sampling rate of 12 KHz. Approximately 47 seconds of data were recorded for each bearing, resulting in 561,152 data points per class. The rotation speed was set to 1800rpm, significantly faster than previous datasets. Moreover, defects were manually injected at the outer race (OR), inner race (IR), and ball, similar to the JNU dataset. However, we established three severity levels (minor, moderate, severe), resulting in ten classes, as detailed in Table 6. This dataset presents a greater challenge than the JNU dataset due to the faults being cracks of the same size but varying depths, leading to more similarity in the features of different categories. The segmentation of the dataset is similar to that of the JNU dataset. A stride of 28 was set to acquire a larger volume of data. The last quarter of the data was designated as the test set, while the remaining data were randomly divided into training and validation sets at a ratio of 0.8:0.2. Consequently, the training set, validation set, and test set comprise 113920, 28480, and 46732 samples, respectively. Lastly, the SNR levels were set at -10dB, -8dB, -6dB, -4dB. 4.4.2. Classification results The classification results are presented in Table 7. Several observations observations are as follows. Firstly, ClassBD outperforms its competitors in terms of both F1 score and FPR across all noise conditions. The F1 score of ClassBD exceeds 96%, compared to the second-best method, EWSNet, which achieves an average F1 score of 92.75%. Secondly, it is noteworthy that even under severe noise conditions (-10dB), ClassBD maintains an F1 score of 93%, indicating its robust anti-noise performance. Under these conditions, the performance of all other methods, except EWSNet, degrades significantly. This phenomenon highlights the superiority of ClassBD. Furthermore, we reduce the dimensions of the last layer features of all methods to 2D space using t-SNE [81] under -10dB noise. The results are depicted in Figure 5. Overall, the features of the five methods can generate distinct clusters, with the exception of TFN, which aligns with their categorical performance. However, all methods exhibit some degree of misclassification. For instance, the red points (OR3) of DRSN and EWSNet appear within the blue clusters (B3), indicating that some instances of OR3 are misclassified as B3. Similarly, in GTFENet and WaveleKernelNet, some green points (IR3) overlap with the IR2 and B3 clusters. In contrast, ClassBD only has a few outliers that are misclassified into other clusters, demonstrating its superior feature extraction ability under high noise conditions. In summary, based on the performance across three datasets, we illustrate the average F1 scores of all methods in Figure 6. The results indicate that our method, ClassBD, outperforms other baseline methods on all datasets, with average F1 scores exceeding 90%. This underscores the robust anti-noise performance of ClassBD. Authors:
(1) Jing-Xiao Liao, Department of Industrial and Systems Engineering, The Hong Kong Polytechnic University, Hong Kong, Special Administrative Region of China and School of Instrumentation Science and Engineering, Harbin Institute of Technology, Harbin, China;
(2) Chao He, School of Mechanical, Electronic and Control Engineering, Beijing Jiaotong University, Beijing, China;
(3) Jipu Li, Department of Industrial and Systems Engineering, The Hong Kong Polytechnic University, Hong Kong, Special Administrative Region of China;
(4) Jinwei Sun, School of Instrumentation Science and Engineering, Harbin Institute of Technology, Harbin, China;
(5) Shiping Zhang (Corresponding author), School of Instrumentation Science and Engineering, Harbin Institute of Technology, Harbin, China;
(6) Xiaoge Zhang (Corresponding author), Department of Industrial and Systems Engineering, The Hong Kong Polytechnic University, Hong Kong, Special Administrative Region of China. This paper is available on arxiv under CC by 4.0 Deed (Attribution 4.0 International) license. Table of Links Abstract and 1. Introduction Abstract and 1. Introduction 2. Preliminaries and 2.1. Blind deconvolution 2. Preliminaries and 2.1. Blind deconvolution 2.2. Quadratic neural networks 2.2. Quadratic neural networks 3. Methodology 3. Methodology 3.1. Time domain quadratic convolutional filter 3.1. Time domain quadratic convolutional filter 3.2. Superiority of cyclic features extraction by QCNN 3.2. Superiority of cyclic features extraction by QCNN 3.3. Frequency domain linear filter with envelope spectrum objective function 3.3. Frequency domain linear filter with envelope spectrum objective function 3.4. Integral optimization with uncertainty-aware weighing scheme 3.4. Integral optimization with uncertainty-aware weighing scheme 4. Computational experiments 4.1. Experimental configurations 4.1. Experimental configurations 4.2. Case study 1: PU dataset 4.2. Case study 1: PU dataset 4.3. Case study 2: JNU dataset 4.3. Case study 2: JNU dataset 4.4. Case study 3: HIT dataset 4.4. Case study 3: HIT dataset 5. Computational experiments 5.1. Comparison of BD methods 5.1. Comparison of BD methods 5.2. Classification results on various noise conditions 5.2. Classification results on various noise conditions 5.3. Employing ClassBD to deep learning classifiers 5.3. Employing ClassBD to deep learning classifiers 5.4. Employing ClassBD to machine learning classifiers 5.4. Employing ClassBD to machine learning classifiers 5.5. Feature extraction ability of quadratic and conventional networks 5.5. Feature extraction ability of quadratic and conventional networks 5.6. Comparison of ClassBD filters 5.6. Comparison of ClassBD filters 6. Conclusions 6. Conclusions Appendix and References Appendix and References 4.4.1. Dataset description 4.4.1. Dataset description 4.4.1. Dataset description 4.4.1. Dataset description The Harbin Institute of Technology (HIT) dataset was collected by our team. The data acquisition for faulty bearings was conducted at the MIIT Key Laboratory of Aerospace Bearing Technology and Equipment at HIT. The bearing test rig and faulty bearings are illustrated in Figure 4. This test rig follows the Chinese standard rolling bearing measuring method (GB/T 32333-2015). The test bearings used were HC7003 angular contact ball bearings, typically employed in high-speed rotating machines. For signal collection, an acceleration sensor was directly attached to the bearing, and the NI USB-6002 device was used at a sampling rate of 12 KHz. Approximately 47 seconds of data were recorded for each bearing, resulting in 561,152 data points per class. The rotation speed was set to 1800rpm, significantly faster than previous datasets. Moreover, defects were manually injected at the outer race (OR), inner race (IR), and ball, similar to the JNU dataset. However, we established three severity levels (minor, moderate, severe), resulting in ten classes, as detailed in Table 6. This dataset presents a greater challenge than the JNU dataset due to the faults being cracks of the same size but varying depths, leading to more similarity in the features of different categories. The segmentation of the dataset is similar to that of the JNU dataset. A stride of 28 was set to acquire a larger volume of data. The last quarter of the data was designated as the test set, while the remaining data were randomly divided into training and validation sets at a ratio of 0.8:0.2. Consequently, the training set, validation set, and test set comprise 113920, 28480, and 46732 samples, respectively. Lastly, the SNR levels were set at -10dB, -8dB, -6dB, -4dB. 4.4.2. Classification results 4.4.2. Classification results 4.4.2. Classification results The classification results are presented in Table 7. Several observations observations are as follows. Firstly, ClassBD outperforms its competitors in terms of both F1 score and FPR across all noise conditions. The F1 score of ClassBD exceeds 96%, compared to the second-best method, EWSNet, which achieves an average F1 score of 92.75%. Secondly, it is noteworthy that even under severe noise conditions (-10dB), ClassBD maintains an F1 score of 93%, indicating its robust anti-noise performance. Under these conditions, the performance of all other methods, except EWSNet, degrades significantly. This phenomenon highlights the superiority of ClassBD. Furthermore, we reduce the dimensions of the last layer features of all methods to 2D space using t-SNE [81] under -10dB noise. The results are depicted in Figure 5. Overall, the features of the five methods can generate distinct clusters, with the exception of TFN, which aligns with their categorical performance. However, all methods exhibit some degree of misclassification. For instance, the red points (OR3) of DRSN and EWSNet appear within the blue clusters (B3), indicating that some instances of OR3 are misclassified as B3. Similarly, in GTFENet and WaveleKernelNet, some green points (IR3) overlap with the IR2 and B3 clusters. In contrast, ClassBD only has a few outliers that are misclassified into other clusters, demonstrating its superior feature extraction ability under high noise conditions. In summary, based on the performance across three datasets, we illustrate the average F1 scores of all methods in Figure 6. The results indicate that our method, ClassBD, outperforms other baseline methods on all datasets, with average F1 scores exceeding 90%. This underscores the robust anti-noise performance of ClassBD. Authors: (1) Jing-Xiao Liao, Department of Industrial and Systems Engineering, The Hong Kong Polytechnic University, Hong Kong, Special Administrative Region of China and School of Instrumentation Science and Engineering, Harbin Institute of Technology, Harbin, China; (2) Chao He, School of Mechanical, Electronic and Control Engineering, Beijing Jiaotong University, Beijing, China; (3) Jipu Li, Department of Industrial and Systems Engineering, The Hong Kong Polytechnic University, Hong Kong, Special Administrative Region of China; (4) Jinwei Sun, School of Instrumentation Science and Engineering, Harbin Institute of Technology, Harbin, China; (5) Shiping Zhang (Corresponding author), School of Instrumentation Science and Engineering, Harbin Institute of Technology, Harbin, China; (6) Xiaoge Zhang (Corresponding author), Department of Industrial and Systems Engineering, The Hong Kong Polytechnic University, Hong Kong, Special Administrative Region of China. Authors: Authors: (1) Jing-Xiao Liao, Department of Industrial and Systems Engineering, The Hong Kong Polytechnic University, Hong Kong, Special Administrative Region of China and School of Instrumentation Science and Engineering, Harbin Institute of Technology, Harbin, China; (2) Chao He, School of Mechanical, Electronic and Control Engineering, Beijing Jiaotong University, Beijing, China; (3) Jipu Li, Department of Industrial and Systems Engineering, The Hong Kong Polytechnic University, Hong Kong, Special Administrative Region of China; (4) Jinwei Sun, School of Instrumentation Science and Engineering, Harbin Institute of Technology, Harbin, China; (5) Shiping Zhang (Corresponding author), School of Instrumentation Science and Engineering, Harbin Institute of Technology, Harbin, China; (6) Xiaoge Zhang (Corresponding author), Department of Industrial and Systems Engineering, The Hong Kong Polytechnic University, Hong Kong, Special Administrative Region of China. This paper is available on arxiv under CC by 4.0 Deed (Attribution 4.0 International) license. This paper is available on arxiv under CC by 4.0 Deed (Attribution 4.0 International) license. available on arxiv

ClassBD Achieves Exceptional Anti-Noise Performance on HIT Dataset with F1 Score Above 96%

You Don't Need Limitless' NZT to Achieve Superhuman Mental Abilities

Не се нуждаете от NZT на Limitless, за да постигнете свръхчовешки умствени способности

K dosažení nadlidských mentálních schopností nepotřebujete neomezené NZT

No necesitas NZT de Limitless para lograr habilidades mentales sobrehumanas

Non hai bisogno dell'NZT di Limitless per ottenere abilità mentali sovrumane

超人的な精神能力を獲得するのにリミットレスのNZTは必要ありません

თქვენ არ გჭირდებათ Limitless' NZT ზეადამიანური გონებრივი შესაძლებლობების მისაღწევად

Ga o Hloke Limitless' NZT go Fihlelela Bokgoni bja Menagano bjo bo Feteletšego bja Motho

Você não precisa do NZT Limitless para obter habilidades mentais sobre-humanas

Manan Necesitankichu Limitless' NZT nisqataqa, aswan runamanta yuyay atiykunata aypanaykipaq

Вам не нужен NZT от Limitless, чтобы достичь сверхчеловеческих умственных способностей

Hindi Mo Kailangan ang Walang Hanggan' NZT para Makamit ang Superhuman Mental Abilities

Awudingi i-NZT Engenamkhawulo ukuze Uzuze Amakhono Engqondo Angaphezu Kwabantu

Table of Links Abstract and 1. Introduction 2. Preliminaries and 2.1. Blind deconvolution 2.2. Quadratic neural networks 3. Methodology 3.1. Time domain quadratic convolutional filter 3.2. Superiority of cyclic features extraction by QCNN 3.3. Frequency domain linear filter with envelope spectrum objective function 3.4. Integral optimization with uncertainty-aware weighing scheme 4. Computational experiments 4.1. Experimental configurations 4.2. Case study 1: PU dataset 4.3. Case study 2: JNU dataset 4.4. Case study 3: HIT dataset 5. Computational experiments 5.1. Comparison of BD methods 5.2. Classification results on various noise conditions 5.3. Employing ClassBD to deep learning classifiers 5.4. Employing ClassBD to machine learning classifiers 5.5. Feature extraction ability of quadratic and conventional networks 5.6. Comparison of ClassBD filters 6. Conclusions Appendix and References 4.3. Case study 2: JNU dataset 4.3.1. Dataset description This dataset was provided by Jiangnan University (JNU) [80]. Two types of roller bearings were artificially injected with inner race defects (bearing N205), outer race defects and ball defects (bearing NU205) by a wire-cutting machine. Three different rotation speeds (600rpm, 800rpm, 1000rpm) were implemented to test the bearings. The sampling rate was set to 50KHz, resulting in signal segments of 20 seconds each. This dataset is one of the more difficult datasets in the field of bearing fault diagnosis [71]. To facilitate varying rotation speed classification, we organized the data into ten classes. These classes encompass three fault types across the three different speeds, along with a healthy class. Furthermore, compared to the PU dataset, the JNU dataset is characterized by a limited data volume. Consequently, we adopted an overlapping strategy to extract signal segments, with a stride of 100. The final dataset configuration involved allocating the last 25% of the data as the test set. Prior to this, the remaining data were randomly partitioned into training and validation sets, maintaining an 80:20 split ratio. In summary, the dataset statistics are as follows: 54,064 samples in the training set, 13,517 samples in the validation set, and 22,269 samples in the test set. Also, we introduce four SNR levels: -10 dB, -8 dB, -6 dB, and -4 dB for evaluation. 4.3.2. Classification results The classification results, as summarized in Table 5, yield valuable insights. Notably, the ClassBD model consistently outperforms its competitors across various noisy conditions. Overall, The ClassBD achieves an impressive F1 score exceeding 96%, surpassing the second-best method, EWSNet, which attains an average F1 score of 92.75%. To be specific, even in the presence of severe noise (at -10 dB), our method maintains a commendable 93% F1 score. This remarkable performance underscores the ClassBD’s excellent anti-noise capabilities. In comparison, under the same noisy conditions, other methods, excluding EWSNet, experience significant degradation. At last, the JNU dataset experiment substantiates that our approach effectively realizes bearing fault diagnosis across varying rotational speeds, even in challenging high-noise environments. Authors:
(1) Jing-Xiao Liao, Department of Industrial and Systems Engineering, The Hong Kong Polytechnic University, Hong Kong, Special Administrative Region of China and School of Instrumentation Science and Engineering, Harbin Institute of Technology, Harbin, China;
(2) Chao He, School of Mechanical, Electronic and Control Engineering, Beijing Jiaotong University, Beijing, China;
(3) Jipu Li, Department of Industrial and Systems Engineering, The Hong Kong Polytechnic University, Hong Kong, Special Administrative Region of China;
(4) Jinwei Sun, School of Instrumentation Science and Engineering, Harbin Institute of Technology, Harbin, China;
(5) Shiping Zhang (Corresponding author), School of Instrumentation Science and Engineering, Harbin Institute of Technology, Harbin, China;
(6) Xiaoge Zhang (Corresponding author), Department of Industrial and Systems Engineering, The Hong Kong Polytechnic University, Hong Kong, Special Administrative Region of China. This paper is available on arxiv under CC by 4.0 Deed (Attribution 4.0 International) license. Table of Links Abstract and 1. Introduction Abstract and 1. Introduction 2. Preliminaries and 2.1. Blind deconvolution 2. Preliminaries and 2.1. Blind deconvolution 2.2. Quadratic neural networks 2.2. Quadratic neural networks 3. Methodology 3. Methodology 3.1. Time domain quadratic convolutional filter 3.1. Time domain quadratic convolutional filter 3.2. Superiority of cyclic features extraction by QCNN 3.2. Superiority of cyclic features extraction by QCNN 3.3. Frequency domain linear filter with envelope spectrum objective function 3.3. Frequency domain linear filter with envelope spectrum objective function 3.4. Integral optimization with uncertainty-aware weighing scheme 3.4. Integral optimization with uncertainty-aware weighing scheme 4. Computational experiments 4.1. Experimental configurations 4.1. Experimental configurations 4.2. Case study 1: PU dataset 4.2. Case study 1: PU dataset 4.3. Case study 2: JNU dataset 4.3. Case study 2: JNU dataset 4.4. Case study 3: HIT dataset 4.4. Case study 3: HIT dataset 5. Computational experiments 5.1. Comparison of BD methods 5.1. Comparison of BD methods 5.2. Classification results on various noise conditions 5.2. Classification results on various noise conditions 5.3. Employing ClassBD to deep learning classifiers 5.3. Employing ClassBD to deep learning classifiers 5.4. Employing ClassBD to machine learning classifiers 5.4. Employing ClassBD to machine learning classifiers 5.5. Feature extraction ability of quadratic and conventional networks 5.5. Feature extraction ability of quadratic and conventional networks 5.6. Comparison of ClassBD filters 5.6. Comparison of ClassBD filters 6. Conclusions 6. Conclusions Appendix and References Appendix and References 4.3. Case study 2: JNU dataset 4.3.1. Dataset description 4.3.1. Dataset description 4.3.1. Dataset description This dataset was provided by Jiangnan University (JNU) [80]. Two types of roller bearings were artificially injected with inner race defects (bearing N205), outer race defects and ball defects (bearing NU205) by a wire-cutting machine. Three different rotation speeds (600rpm, 800rpm, 1000rpm) were implemented to test the bearings. The sampling rate was set to 50KHz, resulting in signal segments of 20 seconds each. This dataset is one of the more difficult datasets in the field of bearing fault diagnosis [71]. To facilitate varying rotation speed classification, we organized the data into ten classes. These classes encompass three fault types across the three different speeds, along with a healthy class. Furthermore, compared to the PU dataset, the JNU dataset is characterized by a limited data volume. Consequently, we adopted an overlapping strategy to extract signal segments, with a stride of 100. The final dataset configuration involved allocating the last 25% of the data as the test set. Prior to this, the remaining data were randomly partitioned into training and validation sets, maintaining an 80:20 split ratio. In summary, the dataset statistics are as follows: 54,064 samples in the training set, 13,517 samples in the validation set, and 22,269 samples in the test set. Also, we introduce four SNR levels: -10 dB, -8 dB, -6 dB, and -4 dB for evaluation. 4.3.2. Classification results 4.3.2. Classification results 4.3.2. Classification results The classification results, as summarized in Table 5, yield valuable insights. Notably, the ClassBD model consistently outperforms its competitors across various noisy conditions. Overall, The ClassBD achieves an impressive F1 score exceeding 96%, surpassing the second-best method, EWSNet, which attains an average F1 score of 92.75%. To be specific, even in the presence of severe noise (at -10 dB), our method maintains a commendable 93% F1 score. This remarkable performance underscores the ClassBD’s excellent anti-noise capabilities. In comparison, under the same noisy conditions, other methods, excluding EWSNet, experience significant degradation. At last, the JNU dataset experiment substantiates that our approach effectively realizes bearing fault diagnosis across varying rotational speeds, even in challenging high-noise environments. Authors: (1) Jing-Xiao Liao, Department of Industrial and Systems Engineering, The Hong Kong Polytechnic University, Hong Kong, Special Administrative Region of China and School of Instrumentation Science and Engineering, Harbin Institute of Technology, Harbin, China; (2) Chao He, School of Mechanical, Electronic and Control Engineering, Beijing Jiaotong University, Beijing, China; (3) Jipu Li, Department of Industrial and Systems Engineering, The Hong Kong Polytechnic University, Hong Kong, Special Administrative Region of China; (4) Jinwei Sun, School of Instrumentation Science and Engineering, Harbin Institute of Technology, Harbin, China; (5) Shiping Zhang (Corresponding author), School of Instrumentation Science and Engineering, Harbin Institute of Technology, Harbin, China; (6) Xiaoge Zhang (Corresponding author), Department of Industrial and Systems Engineering, The Hong Kong Polytechnic University, Hong Kong, Special Administrative Region of China. Authors: Authors: (1) Jing-Xiao Liao, Department of Industrial and Systems Engineering, The Hong Kong Polytechnic University, Hong Kong, Special Administrative Region of China and School of Instrumentation Science and Engineering, Harbin Institute of Technology, Harbin, China; (2) Chao He, School of Mechanical, Electronic and Control Engineering, Beijing Jiaotong University, Beijing, China; (3) Jipu Li, Department of Industrial and Systems Engineering, The Hong Kong Polytechnic University, Hong Kong, Special Administrative Region of China; (4) Jinwei Sun, School of Instrumentation Science and Engineering, Harbin Institute of Technology, Harbin, China; (5) Shiping Zhang (Corresponding author), School of Instrumentation Science and Engineering, Harbin Institute of Technology, Harbin, China; (6) Xiaoge Zhang (Corresponding author), Department of Industrial and Systems Engineering, The Hong Kong Polytechnic University, Hong Kong, Special Administrative Region of China. This paper is available on arxiv under CC by 4.0 Deed (Attribution 4.0 International) license. This paper is available on arxiv under CC by 4.0 Deed (Attribution 4.0 International) license. available on arxiv

How ClassBD Achieved High Accuracy in Bearing Fault Detection Despite High Noise

Stories, News & Mentions about ACHIEVE

Achieve’s Stories on HackerNoon