检索结果-内蒙古大学图书馆

IEEE International Conference on Embedded and Real-Time computing Systems and Applications (RTCSA)

作者： Aditya Bansal Michael Yuhas Arvind Easwaran College of Computing and Data Science Nanyang Technological University Singapore Energy Research Institute @ NTU Interdisciplinary Graduate Program Nanyang Technological University Singapore

ISBN: (数字)9798350387957

ISBN: (纸本)9798350387964

Out-of-distribution (OOD) detectors can act as safety monitors in embedded cyber-physical systems by identifying samples outside a machine learning model’s training distribution to prevent potentially unsafe actions. However, OOD detectors are often implemented using deep neural networks, which makes it difficult to meet real-time deadlines on embedded systems with memory and power constraints. We consider the class of variational autoencoder (VAE) based OOD detectors where OOD detection is performed in latent space, and apply quantization, pruning, and knowledge distillation. These techniques have been explored for other deep models, but no work has considered their combined effect on latent space OOD detection. While these techniques increase the VAE’s test loss, this does not correspond to a proportional decrease in OOD detection performance and we leverage this to develop lean OOD detectors capable of real-time inference on embedded CPUs and GPUs. We propose a design methodology that combines all three compression techniques and yields a significant decrease in memory and execution time while maintaining AUROC for a given OOD detector. We demonstrate this methodology with two existing OOD detectors on a Jetson Nano and reduce GPU and CPU inference time by 20% and 28% respectively while keeping AUROC within 5% of the baseline.

关键词： Training Quantization (signal) Embedded systems Memory management Graphics processing units Detectors Machine learning Real-time systems Safety Monitoring

来源：评论

学校读者我要写书评

暂无评论

MILPdM: A Predictive Maintenance Architecture for the Military Domain 22

MILPdM: A Predictive Maintenance Architecture for the Milita...

引用

22nd IEEE International Conference on Machine Learning and Applications, ICMLA 2023

作者： Dalzochio, Jovani Kunst, Rafael Barbosa, Jorge Pignaton De Freitas, Edison Binotto, Alecio University of Vale Do Rio Dos Sinos Graduate Program in Applied Computing São Leopoldo Brazil Institute of Informatics Federal University of Rio Grande Do sui Porto Alegre Brazil Ibm Consulting Asset and Edge Optimization Munich Germany

ISBN: (纸本)9798350345346

Equipment monitoring for failure prediction is receiving attention from different sectors of society, such as industry, healthcare, and defense. In the defense domain, assets like military vehicles generate data that one can use to identify behavior changes and anticipate possible real-time failures, avoiding unnecessary maintenance interventions. Failure anticipation is crucial, as assets operated in the military domain perform critical tasks in which unexpected equipment failures result in high material and human costs. Approaches found in the literature typically deal with failure generation, aiming at analyzing the equipment's behavior. This paper proposes a broader approach called MILPdM. This proposal is a failure prediction architecture covering the whole failure prediction and predictive maintenance procedures. We evaluate MILPdM architecture by analyzing an engine-failure scenario where we train models to predict time series by collecting vibration data that describes the degradation of the vehicle's health. Considering the implementation of an LSTM-based neural network and Random Forest, the acquired results lead to a root mean square error of 0.15015 in the best case, which allows to predict the failure status two minutes in advance with only 3 hours of data history. This result shows that MILPdM is capable of anticipating failures with high assertiveness. © 2023 IEEE.

关键词： IoBT machine learning neural networks predictive maintenance random forest

来源：评论

学校读者我要写书评

暂无评论

Using Decision Trees to Accelerate the H.266/VVC-to-AV1 Video Transcoding

Using Decision Trees to Accelerate the H.266/VVC-to-AV1 Vide...

引用

Brazilian Symposium on Computer Graphics and Image Processing (SIBGRAPI)

作者： Caroline Camargo Alex Borges Bruno Zatt Marcelo Porto Guilherme Correa Video Technology Research Group - ViTech Graduate Program in Computing - PPGC Federal University of Pelotas - UFPel Pelotas Brazil

H.266/NVC is the current state-of-the-art video coding format, launched in 2020. Despite its high capability to compress video, including ultra-high definition content, royalty costs may prevent its wide adoption. The AOMedia Video 1 (AV1) format is an alternative with competitive coding efficiency, besides being a royalty-free format. However, migrating legacy content from one format to another is a costly task, which requires long processing times. This work presents a solution for accelerating the H.266/NVC-to-AV1 trans coding based on machine learning. Twelve decision tree models trained with data gathered during the H.266NVC decoding and the AV1 encoding processes are proposed and implemented in the libaom reference software, leading to a complexity reduction of 12.60 % at the cost of coding efficiency losses of 1.81 % on average. To the best of the authors' knowledge, this is the first H.266NVC-to-AV1 trans coding acceleration solution published in the literature.

关键词：

来源：评论

学校读者我要写书评

暂无评论

ARTERIAL: A Natural Language Processing Model for Prevention of Information Leakage from Electronic Health Records

ARTERIAL: A Natural Language Processing Model for Prevention...

引用

Brazilian Symposium on computing System Engineering (SBESC)

作者： Guilherme Goldschmidt Felipe André Zeiser Rodrigo Da Rosa Righi Cristiano André Da Costa Graduate Program in Applied Computing Universidade do Vale do Rio dos Sinos São Leopoldo Brazil

Over the past decade, there has been a steady increase in health security breaches. Therefore, healthcare organizations must protect their sensitive information such as test results, diagnoses, prescriptions, research, and customer personal information. A leak of sensitive data can result in significant economic loss and damage to the organization’s image. In this sense, Data Leakage Prevention (DLP) systems can help to identify, monitor, protect, and reduce the risks of leaking sensitive data. However, state-of-the-art DLP solutions only use signature comparisons and static comparisons. Therefore, we propose to develop the ARTERIAL model based on Natural Language Processing (NLP), Entity Recognition (NER), and Artificial Neural Networks (ANN) to be more assertive in extracting information and recognizing entities from Electronic Health Records (EHR). Different from the current literature, ARTERIAL considers semantic features present in the EHR. Three approaches were implemented and tested, two based on ANN and the following based on machine learning algorithms. As a result, the approach taken in its implementation using a machine learning algorithm reached 98.0% of Precision, 86.0% of Recall, and 91.0% of F1-Score.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Efficient Hardware Design for the VVC Affine Motion Compensation Exploiting Multiple Constant Multiplication

Efficient Hardware Design for the VVC Affine Motion Compensa...

引用

IEEE Computer Society Annual Symposium on VLSI

作者： Marcello M. Muñoz Denis Maass Murilo Perleberg Luciano Agostini Marcelo Porto Graduate Program in Computing - PPGC Video Technology Research Group - ViTech Federal University of Pelotas - UFPel Pelotas Brazil

The Affine Motion Estimation (AME) is a new and high-complexity task of the Versatile Video Coding (VVC) standard. The AME requires the Affine Motion Compensation (MC) to be performed for 4$\times$ 4 subblocks, where one among 156-tap interpolation filters was adopted to interpolate each sample of the 4$\times$ 4 subblock according to the motion vector relative to this subblock. This work presents two dedicated hardware implementations for the Affine MC of the VVC standard, the first focusing on the reduction of power dissipation and the second on the area requirement. The ASIC synthesis results of these architectures for TSMC 40nm standard cells show an area requirement of 54. 43k gates and power dissipation of 12. 8mW for the power efficient variant, while for the hardware efficient, the area requirement is 21. 91k gates and power dissipation of 14.41mW.

关键词：

来源：评论

学校读者我要写书评

暂无评论

H.264-to-AV1 Video Transcoding Acceleration Based on Lightweight Machine Learning

H.264-to-AV1 Video Transcoding Acceleration Based on Lightwe...

引用

IEEE International Symposium on Circuits and Systems (ISCAS)

作者： Alex Borges Marcelo Porto Bruno Zatt Guilherme Correa Video Technology Research Group (ViTech) Graduate Program in Computing (PPGC) Federal University of Pelotas (UFPel) Pelotas Brazil

Video streaming platforms have been using the H.264/AVC standard for a long time, even though it was released almost 20 years ago and much more efficient codecs are currently available. The AOMedia Video 1 (AV1) format is an alternative with significant coding efficiency gains in comparison to H.264/AVC, besides being a royalty-free format. However, migrating legacy content from older to newer formats is a costly task, which requires long processing times. This work presents a solution for accelerating the H.264-to-AV1 transcoder based on machine learning. Sixteen decision tree models trained with data gathered during the H.264/AVC decoding and the AV1 encoding processes are proposed and implemented in the libaom reference software, leading to a complexity reduction of 18.96% at the cost of coding efficiency losses of 2.85% on average. To the best of the authors' knowledge, this is the first H.264-to-AV1 transcoding acceleration solution published in the literature.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Investigating Cases and Clusters-Based Reuse Policies for Card-Playing Agents 5

Investigating Cases and Clusters-Based Reuse Policies for Ca...

引用

5th Annual IEEE Conference on Games, CoG 2023

作者： Paulus, Gustavo B. Vargas, Daniel P. Assuncao, Joaquim V. C. Silva, Luis A. L. Federal Institute of Education Science and Technology of Rio Grande do Sul Ibirubá Brazil Federal University of Santa Maria Graduate Program in Computer Science Santa Maria Brazil Federal University of Santa Maria Applied Computing Department Santa Maria Brazil

ISBN: (纸本)9798350322774

Computer games continue to present challenging experimental fields for developing Artificial Intelligence (AI) models. With Case-Based Reasoning and Clustering, this work proposes novel cases and clusters-based reuse criteria for implementing card-playing agents with diversified playing skills. Using the game of Truco, a common game in South America, we detail how game actions are reused from past cases selected as query answers for given game problems. In doing so, the majority rule, the probability-based lottery, the probability of victory, and the number of points won reuse policies are used to select a cluster of cases. Then these policies are also used to select game actions from the cases within the selected cluster. Investigating the combined exploration of reuse policies, experiments of different natures evaluate the performance of implemented Truco bots disputing matches against each other. Players in these tournaments are equipped with varied policies and use case bases constructed differently. © 2023 IEEE.

关键词： Case based reasoning

来源：评论

学校读者我要写书评

暂无评论

Protein Structure Prediction Using Dynamic Speciation Evolutionary Algorithm with Aggregated Problem Information

Protein Structure Prediction Using Dynamic Speciation Evolut...

引用

IEEE Symposium on Computational Intelligence and Bioinformatics and Computational Biology (CIBCB)

作者： Rafael Stubs Parpinelli Nicholas Wojeicchowski Nilcimar Neitzel Will Graduate Program in Applied Computing State University of Santa Catarina Joinville Brazil Computer Science Department State University of Santa Catarina Joinville Brazil

ISBN: (数字)9798350356632

ISBN: (纸本)9798350356649

Protein structure prediction in three dimensions represents a fundamental challenge in Structural Bioinformatics. Leveraging problem-specific information such as fragment insertion, secondary structure, and contact maps can significantly enhance the exploration of the search space. In this study, an evolutionary algorithm is introduced, which incorporates such problem information for protein structure prediction. The proposed method employs a dynamic speciation technique alongside fragment insertion to foster population diversity. To ensure a rich variety of fragments, a fragment library is constructed using the Rosetta Quota protocol. Additionally, information from contact maps and secondary structure is integrated into two selection strategies to facilitate a more thorough exploration of the conformational search space. The results of an experimental evaluation involving 9 proteins are presented, demonstrating competitive performance compared to existing literature. Evaluation metrics include RMSD, GDT, and processing time.

关键词： Proteins Measurement Visualization Protocols Heuristic algorithms Evolutionary computation Search problems Libraries Space exploration Bioinformatics

来源：评论

学校读者我要写书评

暂无评论

An Exploratory Study to Assess the Usability of a Groupware with Multi-Agent Systems

An Exploratory Study to Assess the Usability of a Groupware ...

引用

Frontiers in Education (FIE) Conference

作者： Joethe Moraes de Carvalho José Francisco de Magalhães Netto Post-Graduate Program in Informatics (PPGI) Institute of Computing (ICOMP) Federal University of Amazonas (UFAM) CMC-Federal Institute of Amazonas (IFAM) Manaus AM Brazil Post-Graduate Program in Informatics (PPGI) Institute of Computing (ICOMP) Federal University of Amazonas (UFAM) Manaus AM Brazil

This Research Full Paper presents a pedagogical approach to evaluate a virtual platform for teaching robotics at a distance. The virtual environment was developed using the principles of Collaborative Learning. This virtual platform, called groupware, should facilitate interaction between those involved in the teaching and learning process, whether the student or the teacher. In this research, we will use Educational Robotics as a mediating tool in the teaching and learning process, because it has become a powerful tool that has stimulated the acquisition of knowledge in computing. In the groupware used in this research, Multi-Agent Systems were used to collaborate with teachers in the management of activities. The purpose of this study was to evaluate the functionality and usability of this virtual environment. A qualitative, exploratory research was used and to assess the usability of this virtual environment, the usability evaluation techniques were used with the Syetem Usabily Scale metric, and the technique of evaluation of feelings through EmoCards. Ten computer teachers participated in this research, with training and specializations in the areas of computing. Collaborative Learning was used as a focus for solving the tasks proposed for this research. The results showed that the participants evaluated the groupware with great potential to collaborate in robotics teaching and with content appropriate to the students' initial level of learning. The teachers' perceptions regarding the usability of the evaluated groupware revealed that it is favorable in the teaching of robotics and allowed an analysis of the importance of collaborative learning in the educational environment, directly impacting their ways of conducting learning.

关键词：

来源：评论

学校读者我要写书评

暂无评论

A Comprehensive Technique Based on Machine Learning for Device and Circuit Modeling of Gate-All-Around Nanosheet Transistors

IEEE Open Journal of Nanotechnology

引用

IEEE Open Journal of Nanotechnology 2023年 4卷 181-194页

作者： Butola, Rajat Li, Yiming Kola, Sekhar Reddy National Yang Ming Chiao Tung University Parallel and Scientific Computing Laboratory Electrical Engineering and Computer Science International Graduate Program Hsinchu300093 Taiwan National Yang Ming Chiao Tung University Institute of Communications Engineering the Institute of Biomedical Engineering the Department of Electronics and Electrical Engineering the Institute of Pioneer Semiconductor Innovation and the Institute of Artificial Intelligence Innovation Hsinchu300093 Taiwan

Machine learning (ML) is poised to play an important part in advancing the predicting capability in semiconductor device compact modeling domain. One major advantage of ML-based compact modeling is its ability to capture complex relationships and patterns in large datasets. Therefore, in this paper a novel design scheme based on dynamically adaptive neural network (DANN) is proposed to develop fast and accurate compact model (CM). This framework constitutes a powerful yet computationally efficient methodology and exhibits emergent dynamic behaviors. This paper demonstrates that the compact model based on ML can be designed to replicate the performance of conventional compact model for nanodevices. For this work, gate-all-around (GAA) nanosheet (NS) device characteristics are comprehensively analyzed for process variability sources using the proposed model. The device geometry parameters such as channel length, nanosheet width and nanosheet thickness are fed as input features to the DANN model. The adaptive neural network learns dynamically by updating weights of the model in accordance with the input features and achieves accurate neural weight convergence. The proposed model predicted the electrical characteristics of NS devices with less than 1% error rate. The model is also implemented and validated for the simulations of digital circuit designs such as inverter, and logic gates. ©;2023 The Authors.

关键词： Circuit simulation

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：