检索结果-内蒙古大学图书馆

Constraint-Guided Learning of data-driven Health Indicator Models: An Application on the Pronostia Bearing dataset

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Tefera, Yonas Van Baelen, Quinten Meire, Maarten Luca, Stijn Karsmakers, Peter KU Leuven Dept. of Computer Science ADVISE-DTAI Kleinhoefstraat 4 GeelB-2440 Belgium Leuven.AI - KU Leuven Institute for AI Belgium Flanders Make @ KU Leuven Belgium Addis Ababa University School of Electrical and Computer Engineering Addis Ababa Ethiopia Ghent University Department of Data Analysis and Mathematical Modelling Coupure Links 653 Gent9000 Belgium

This paper presents a constraint-guided deep learning framework to develop physically consistent health indicators in bearing prognostics and health management. Conventional data-driven approaches often lack physical plausibility, while physics-based models are limited by incomplete knowledge of complex systems. To address this, we integrate domain knowledge into deep learning models via constraints, ensuring monotonicity, bounding output ranges between 1 and 0 (representing healthy to failed states respectively), and maintaining consistency between signal energy trends and health indicator estimates. This eliminates the need for complex loss term balancing to incorporate domain knowledge. We implement a constraint-guided gradient descent optimization within an autoencoder architecture, creating a constrained autoencoder. However, the framework is flexible and can be applied to other architectures as well. Using time-frequency representations of accelerometer signals from the Pronostia dataset, the constrained model generates more accurate and reliable representations of bearing health compared to conventional methods. It produces smoother degradation profiles that align with the expected physical behavior. Model performance is assessed using three metrics: trendability, robustness, and consistency. When compared to a conventional baseline model, the constrained model shows significant improvement in all three metrics. Another baseline incorporated the monotonicity behavior directly into the loss function using a soft-ranking approach. While this approach outperforms the constrained model in trendability, due to its explicit monotonicity enforcement, the constrained model performed better in robustness and consistency, providing stable and interpretable health indicator estimates over time. The results of the ablation study confirm that the monotonicity constraint enhances trendability, the boundary constraint ensures consistency, and the energy-health indicator con

关键词： Electronic health record

Enhancing Natural Language Processing in Multilingual Chatbots for Cross-Cultural Communication

学校读者我要写书评

暂无评论

Enhancing Natural Language Processing in Multilingual Chatbo...

Intelligent Communication Technologies and Virtual Mobile Networks (ICICV), International Conference on

作者： Myagmarsuren Orosoo Indrajit Goswami Fredrick Ruban Alphonse Gulnaz Fatma Manikandan Rengarajan B Kiran Bala Mongolian National University of Education General Management Strategy and HRM Alliance School of Business Alliance University Bengaluru Department of Languages Institute of Finance and International Management Bengaluru Dept. of English Jazan University Jazan Saudi Arabia Vel Tech Rangarajan Dr. Sagunthala R&D Institute of Science and Technology Chennai Tamil Nadu India Department of Artificial Intelligence and Data Science K.Ramakrishnan college of Engineering Trichy

In order to promote smooth cross-cultural communication, this research focuses on improving Natural Language Processing (NLP) features within multilingual Chabots. The need for efficient communication across various linguistic and cultural barriers has grown as the world becomes more interconnected. In this context, multilingual Chabot's are essential tools that facilitate easy communication between users of various cultural backgrounds. In order to better understand and respond to a wide range of linguistic details, idioms, and cultural allusions, the research explores ways to improve NLP algorithms. This research investigates how machine learning techniques can be integrated to improve the Chabot's ability to adapt to various linguistic patterns, enabling it to acquire new language skills over time. By implementing an extensive and flexible approach, the proposed approach seeks to transform Natural Language Processing in Multilingual Chatbots for Cross-Cultural Communication. In order to identify language-specific patterns and specifications, the methodology starts with the collection of a diverse dataset that is representative of linguistic variations and cultural contexts. It then uses advanced linguistic analysis techniques like phonetic analysis. The next step is cultural context modelling, which entails building a database enhanced with cultural allusions and language variations unique to a given context that are obtained from various sources. Transfer learning, reinforcement learning for dialogue, and modern algorithms such as Multilingual Unsupervised and Supervised Embeddings (MUSE) for cross-lingual embeddings are all included in machine learning integration. In order to obtain information about language precision and cultural sensitivity, the methodology incorporates interactive surveys and sentiment analysis into a strong user feedback mechanism. The proposed model achieved 98% performance.

关键词：

Crack Catcher AI — Enabling Smart Fracture Mechanics Approach for Damage Control in Thin Silicon Wafers for 3D Semiconductor Integrated Devices/Packages

学校读者我要写书评

暂无评论

Crack Catcher AI — Enabling Smart Fracture Mechanics Approa...

Annual International Symposium on Reliability Physics

作者： Gabriel Preston Greer Gonzolez Andrea S. Karnyoto Fitya S. Mozar Matthew M. Henry Mahmud Isnan Kuncahyo S. Nugroho Henry Candra Muhamad Doris Dianing N. N. Putri Tyas K Sari Bens Pardamean Endang Djuana Derrric Speaks Arief S. Budiman Oregon Renewable Energy Center (OREC) Oregon Institute of Technology (Oregon Tech) Klamath Falls OR USA Dept. of Manufacturing & Mechanical Eng. & Tech. (MMET) Oregon Institute of Technology Klamath Falls OR USA Bioinformatics & Data Science Research Center (BDSRC) Bina Nusantara University Jakarta Indonesia Computer Sci. Dept. BINUS Grad. Program - Master of Computer Sci. Bina Nusantara University Jakarta Indonesia Computer Science Department School of Computer Science Bina Nusantara University Jakarta Indonesia Center for Artificial Intelligence and Advanced Technology (CAPTIVATE) Universitas TRISAKTI Jakarta Indonesia Department of Electrical Engineering Faculty of Industrial Technology Universitas TRISAKTI Jakarta Indonesia Industrial Eng. Dept. BINUS Grad. Program – Master of Industrial Eng. Bina Nusantara University Jakarta Indonesia

ISBN: (数字)9798331504779

ISBN: (纸本)9798331504786

The importance of silicon cannot be undermined - in photovoltaics (PV) as well as in semiconductor industries. However, silicon is very brittle. Silicon cells/wafers crack easily during manufacturing assembly and/or during device operations. Crack Catcher AI uses novel smart fracture mechanics approach with Artificial Intelligence (AI) methodologies to predict and control crack/damage evolution in thin silicon cells/wafers. This is critical as semiconductor 3D integration technology calls for wafer-to-wafer bonding with utmost alignment accuracy and yield/reliability.

关键词： Semiconductor device measurement Three-dimensional displays Accuracy Semiconductor device reliability Silicon Artificial intelligence Bonding Stress Three-dimensional integrated circuits Stress measurement

RAPID MODEL TRANSFER for MEDICAL IMAGE SEGMENTATION VIA ITERATIVE HUMAN-IN-THE-LOOP UPDATE: From LABELLED PUBLIC to UNLABELLED CLINICAL dataSETS for MULTI-ORGAN SEGMENTATION in CT

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Ma, Wenao Zheng, Shuang Zhang, Lei Zhang, Huimao Dou, Qi Dept. of Computer Science and Engineering The Chinese University of Hong Kong Hong Kong Hong Kong Department of Radiology The First Hospital of Jilin University Jilin China Jilin Provincial Key Laboratory of Medical Imaging & Big Data Jilin China

Despite the remarkable success on medical image analysis with deep learning, it is still under exploration regarding how to rapidly transfer AI models from one dataset to another for clinical applications. This paper presents a novel and generic human-in-the-loop scheme for efficiently transferring a segmentation model from a small-scale labelled dataset to a larger-scale unlabelled dataset for multi-organ segmentation in CT. To achieve this, we propose to use an igniter network which can learn from a small-scale labelled dataset and generate coarse annotations to start the process of human-machine interaction. Then, we use a sustainer network for our largerscale dataset, and iteratively updated it on the new annotated data. Moreover, we propose a flexible labelling strategy for the annotator to reduce the initial annotation workload. The model performance and the time cost of annotation in each subject evaluated on our private dataset are reported and analysed. The results show that our scheme can not only improve the performance by 19.7% on Dice, but also expedite the cost time of manuallabelling from 13.87 min to 1.51 min per CT volume during the model transfer, demonstrating the clinical usefulness with promising potentials. © 2022, CC BY.

关键词： Image segmentation

An Empirical Evaluation of NVM-aware File Systems on Intel Optane DC Persistent Memory Modules

学校读者我要写书评

暂无评论

An Empirical Evaluation of NVM-aware File Systems on Intel O...

International Conference on Information Networking

作者： Guangyu Zhu Jaehyun Han Sangjin Lee Yongseok Son CAU Institute of Innovative Talent of Big Data Dept. of Computer Science and Engineering and Chung-Ang University Seoul South Korea School of Computer Science and Engineering Chung-Ang University Seoul South Korea

ISBN: (数字)9781728191010

ISBN: (纸本)9781728191027

Cloud computing arouses the need for large-scale data processing which in turn promises vigorous developments on distributed file systems. The performance of the underlying storage nodes determines the performance of the overall system. Emerging byte-addressable Non-volatile memories (NVM) are promising techniques that can greatly improve storage performance. Researchers have already investigated NVM or NVMaware file systems to take advantage of the characteristics of NVM. However, previous researchers usually perform the studies based on simulations or emulations. In this paper, we provide an empirical evaluation of NVM-aware file systems on the first commercially available byte-addressable NVM (i.e., the Intel Optane DC Persistent Memory Module (DCPMM)) We first evaluate the performance of Ext4, XFS, F2FS, Ext4-DAX, XFSDAX, and NOVA file systems on DCPMM, Optane SSD, and NVMe SSD. Then we compare them in terms of throughput and different access patterns. Second, we observe how remote NUMA node access and device mapper striping affect the performance of DCPMM. We anticipate that the experimental results and performance analysis will provide the implications on various memory and storage systems.

关键词： Performance evaluation Nonvolatile memory File systems Sockets Memory modules Throughput Performance analysis

Achieving Linear Speedup with Network-Independent Learning Rates in Decentralized Stochastic Optimization

学校读者我要写书评

暂无评论

Achieving Linear Speedup with Network-Independent Learning R...

IEEE Conference on Decision and Control

作者： Hao Yuan Sulaiman A. Alghunaim Kun Yuan Center for Machine Learning Research Peking University Beijing P. R. China Dept. Electrical Engr. Kuwait University Safat Kuwait AI for Science Institute Beijing P. R. China National Engineering Labratory for Big Data Analytics and Applications Beijing P. R. China

Decentralized stochastic optimization has become a crucial tool for addressing large-scale machine learning and control problems. In decentralized algorithms, all computing nodes are connected through a network topology, and each node communicates only with its direct neighbors. Decentralized algorithms can significantly reduce communication overhead by eliminating the need for global communication. However, existing research on the linear speedup analysis of decentralized stochastic algorithms is limited to the condition of network-dependent learning rates, which rarely holds in practice since the network connectivity is typically unknown to each node. As a result, it remains an open question whether a linear speedup bound can be achieved using network-independent learning rates. This paper provides an affirmative answer. By utilizing a new analysis framework, we prove that D-SGD and Exact-Diffusion, two representative decentralized stochastic algorithms, can achieve linear speedup with network-independent learning rates. Simulations are provided to validate our theories.

关键词：

DocAsRef: An Empirical Study on Repurposing Reference-based Summary Quality Metrics as Reference-free Metrics

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Bao, Forrest Sheng Tu, Ruixuan Luo, Ge Yang, Yinfei Li, Hebi Qiu, Minghui He, Youbiao Chen, Cen Department of Computer Science Iowa State University AmesIA United States Dept. of Computer Sciences University of Wisconsin–Madison MadisonWI United States Sunnyvale CA United States ByteDance China School of Data Science and Engineering East China Normal University Shanghai China

Automated summary quality assessment falls into two categories: reference-based and reference-free. Reference-based metrics, historically deemed more accurate due to the additional information provided by human-written references, are limited by their reliance on human input. In this paper, we hypothesize that the comparison methodologies used by some reference-based metrics to evaluate a system summary against its corresponding reference can be effectively adapted to assess it against its source document, thereby transforming these metrics into reference-free ones. Experimental results support this hypothesis. After being repurposed reference-freely, the zero-shot BERTScore using the pretrained DeBERTa-large-MNLI model of Copyright © 2022, The Authors. All rights reserved.

关键词： Large dataset

Vyakarana: A colorless green benchmark for syntactic evaluation in indic languages

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Patil, Rajaswa Dhillon, Jasleen Mahurkar, Siddhant Kulkarni, Saumitra Malhotra, Manav Baths, Veeky Cognitive Neuroscience Lab BITS Pilani Goa Campus Dept. of CS & IS BITS Pilani Goa Campus Data Science Institute Columbia University Dept. of Computer Engineering PICT Maharashtra India Dept. of Mathematics BITS Pilani Goa Campus

While there has been significant progress towards developing NLU resources for Indic languages, syntactic evaluation has been relatively less explored. Unlike English, Indic languages have rich morphosyntax, grammatical genders, free linear word-order, and highly inflectional morphology. In this paper, we introduce Vyakarana: a benchmark of Colorless Green sentences in Indic languages for syntactic evaluation of multilingual language models. The benchmark comprises four syntax-related tasks: PoS Tagging, Syntax Tree-dept. Prediction, Grammatical Case Marking, and Subject-Verb Agreement. We use the datasets from the evaluation tasks to probe five multilingual language models of varying architectures for syntax in Indic languages. Due to its prevalence, we also include a code-switching setting in our experiments. Our results show that the tokenlevel and sentence-level representations from the Indic language models (IndicBERT and MuRIL) do not capture the syntax in Indic languages as efficiently as the other highly multilingual language models. Further, our layerwise probing experiments reveal that while mBERT, DistilmBERT, and XLM-R localize the syntax in middle layers, the Indic language models do not show such syntactic localization. © 2021, CC BY.

关键词： Syntactics

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Lee, Ming-Lun Chou, Han-Chang Chen, Yan-Ann Dept. of Computer Science and Engineering Yuan Ze University Taoyuan Taiwan Innovation Center for Big Data and Digital Convergence Yuan Ze University Taoyuan Taiwan

Federated learning is a distributed machine learning framework to collaboratively train a global model without uploading privacy-sensitive data onto a centralized server. Usually, this framework is applied to edge devices such as smartphones, wearable devices, and Internet of Things (IoT) devices which closely collect information from users. However, these devices are mostly battery-powered. The update procedure of federated learning will constantly consume the battery power and the transmission bandwidth. In this work, we propose an update control for federated learning, FedSAUC, by considering the similarity of users’ behaviors (models). At the server side, we exploit clustering algorithms to group devices with similar models. Then we select some representatives for each cluster to update information to train the model. We also implemented a testbed prototyping on edge devices for validating the performance. The experimental results show that this update control will not affect the training accuracy in the long run. © 2025, CC BY-NC-ND.

关键词： Edge computing