检索结果-内蒙古大学图书馆

5th IEEE International Conference on Autonomic Computing and Self-Organizing Systems Companion, ACSOS-C 2024

作者： Vindbjerg, Lukas Koch Esterle, Lukas Aarhus University Department of Electircal and Computer Engineering Denmark Aarhus University Department of Electrical and Computer Engineering DIGIT Denmark

ISBN: (纸本)9798350389760

Autonomous systems rely on artificial intelligence to perform their tasks more effectively. With the increasing complexity of tasks, it is essential to provide a structured way to define tasks. This paper explores a novel approach to extract task definitions with temporal information using generative models. The approach combines the strengths of the T5 and BERT models to understand and sequence tasks. Specifically, the BERT model classifies the temporal relationship between tasks before the T5 model is used to generate the DSL sequence. Furthermore, we present a simple DSL for a household robot use-case and a dataset of natural language commands with temporal information. The model is trained on this dataset and evaluated on its ability to generate accurate and temporally ordered task definitions. © 2024 IEEE.

关键词： Adversarial machine learning

来源：评论

学校读者我要写书评

暂无评论

Enhanced Single Diode Model for Bifacial PV Module Incorporating Partial Shading

Enhanced Single Diode Model for Bifacial PV Module Incorpora...

引用

2023 International Conference on Sustainable Technology and engineering, i-COSTE 2023

作者： Ranasinghe, Githmi Rajapakse, Athula University of Manitoba Department of Electircal and Computer Engineering WinnipegMB Canada

ISBN: (纸本)9798350329711

Bifacial photovoltaic (PV) technology, designed to harness solar energy from both the front and rear surfaces of solar modules, presents a promising avenue for enhancing solar energy capture. While the Single Diode Model (SDM) has been widely validated for monofacial PV modules, its accuracy in the context of bifacial modules remains underexplored. This paper presents an advanced SDM, originally tailored for mono facial solar cells, adapted to accommodate the unique characteristics of bifacial PV modules, including the incorporation of partial shading effects. The paper explores bifacial solar cell technologies, bifacial gain parameters, and the lEC TS 60904-1-2 specification, emphasizing the significance of accurate PV modeling in system simulation. The model's accuracy is validated using manufacturer datasheets, which includes variations in Isc (short-circuit current), Voc (open-circuit voltage), and Pmax (maximum power output) for different bifacial gains. Additionally, a modeling approach for partial shading is presented and validated using available experimental data from previous research. © 2023 IEEE.

关键词： Solar energy

来源：评论

学校读者我要写书评

暂无评论

Effective Activation Scratchpad Management in NPUs for LLMs via Inter- and Intra-Decoder Allocation

Effective Activation Scratchpad Management in NPUs for LLMs ...

引用

2025 IEEE International Conference on Consumer Electronics, ICCE 2025

作者： Seo, Minseok Jeong, Seongho Hyun, Jungi Lee, Hyuk-Jae Nguyen, Xuan Truong Electircal and Computer Engineering Seoul National University Inter-University Semiconductor Research Center Seoul Korea Republic of Electircal and Computer Engineering Seoul National University Seoul Korea Republic of

ISBN: (纸本)9798331521165

To accelerate emerging Large Language Models (LLMs), modern Neural Processing Units (NPUs) typically adopt a scratchpad memory for activations as on-chip memory. However, managing an activation scratchpad in NPUs becomes challenge, as the size of LLMs and their context sequence lengths continue to grow. Suboptimal activation memory management may cause low on-chip memory usage and excessive off-chip memory accesses due to large activation data, leading to significant performance degradation during LLM inference on NPUs. This challenge necessitates careful methods to maximize on-chip memory utilization. To address the problem, this work proposes an effective activation scratchpad management in NPUs for LLMs via inter- and intra-decoder allocation. Specifically, inter-decoder optimization facilitates the sharing of on-chip memory across stacked decoders, and intra-decoder one interleaves on-chip memory usage among operations in a decoder. We evaluate our approach with OPT models. The experimental results demonstrate that the proposed approach achieves 56-79× reduction in memory usage to naïve on-chip memory management methods. © 2025 IEEE.

关键词： Memory management

来源：评论

学校读者我要写书评

暂无评论

Effective Activation Scratchpad Management in NPUs for LLMs via Inter- and Intra-Decoder Allocation

Effective Activation Scratchpad Management in NPUs for LLMs ...

引用

IEEE International Conference on Consumer Electronics (ICCE)

作者： Minseok Seo Seongho Jeong Jungi Hyun Hyuk-Jae Lee Xuan Truong Nguyen Inter-University Semiconductor Research Center Electircal and Computer Engineering Seoul National University Seoul South Korea Electircal and Computer Engineering Seoul National University Seoul South Korea

ISBN: (数字)9798331521165

ISBN: (纸本)9798331521172

To accelerate emerging Large Language Models (LLMs), modern Neural Processing Units (NPUs) typically adopt a scratchpad memory for activations as on-chip memory. However, managing an activation scratchpad in NPUs becomes challenge, as the size of LLMs and their context sequence lengths continue to grow. Suboptimal activation memory management may cause low on-chip memory usage and excessive off-chip memory accesses due to large activation data, leading to significant performance degradation during LLM inference on NPUs. This challenge necessitates careful methods to maximize on-chip memory utilization. To address the problem, this work proposes an effective activation scratchpad management in NPUs for LLMs via inter-and intra-decoder allocation. Specifically, inter-decoder optimization facilitates the sharing of on-chip memory across stacked decoders, and intra-decoder one interleaves on-chip memory usage among operations in a decoder. We evaluate our approach with OPT models. The experimental results demonstrate that the proposed approach achieves 56–79 × reduction in memory usage to naïve on-chip memory management methods.

关键词： Degradation Large language models Memory management System-on-chip Decoding Resource management Optimization Consumer electronics

来源：评论

学校读者我要写书评

暂无评论

Rethinking Mobilevitv2 for Real-Time Semantic Segmentation 4

Rethinking Mobilevitv2 for Real-Time Semantic Segmentation

引用

4th International Conference on computers and Artificial Intelligence Technology, CAIT 2023

作者： Zhang, Zhengbin Xu, Zhenhao Gu, Xingsheng Xiong, Juan East China University of Science and Technology School of Information Science and Engineering Shanghai China University of Shanghai for Science and Technology School of Optical Electircal and Computer Engineering Shanghai China

ISBN: (纸本)9798350326710

Semantic segmentation empowers various real-world applications. Nevertheless, the substantial computational cost, such as O(k2) time complexity associated with the number of tokens in multi-headed self-attention, poses challenges for deploying these models on edge devices with constrained hardware resources. This paper introduces a novel family of backbones designed for real-time semantic segmentation, referred to as Linear and Re-parameter Vision Transformer (LARFormer). Particularly, we introduce a Re-parameter Mobile Block (RMB), which employs three branches during training and a single branch during inference. Furthermore, we introduce Linear Separable Self-Attention(LSSA), which reduces the computational complexity from O(k2) to O(k). Extensive experiments on the ADE20K dataset and Pascal VOC 2012 dataset demonstrate the effectiveness of the proposed LARFormer by achieving a promising trade-off between segmentation accuracy and inference speed. © 2023 IEEE.

关键词： Semantic Segmentation

来源：评论

学校读者我要写书评

暂无评论

Generative Models for Temporal-Based Task Definition

Generative Models for Temporal-Based Task Definition

引用

Autonomic Computing and Self-Organizing Systems Companion (ACSOS-C), IEEE International Conference on

作者： Lukas Koch Vindbjerg Lukas Esterle Department of Electircal and Computer Engineering Aarhus University Denmark Department of Electrical and Computer Engineering & DIGIT Aarhus University Denmark

ISBN: (数字)9798350389760

ISBN: (纸本)9798350389777

关键词： Training Analytical models Accuracy Computational modeling Training data Data models DSL Robots Testing Overfitting

来源：评论

学校读者我要写书评

暂无评论

Enhanced Single Diode Model for Bifacial PV Module Incorporating Partial Shading

Enhanced Single Diode Model for Bifacial PV Module Incorpora...

引用

Sustainable Technology and engineering (i-COSTE), International Conference on

作者： Githmi Ranasinghe Athula Rajapakse Department of Electircal and Computer Engineering University of Manitoba Winnipeg Manitoba Canada

Bifacial photovoltaic (PV) technology, designed to harness solar energy from both the front and rear surfaces of solar modules, presents a promising avenue for enhancing solar energy capture. While the Single Diode Model (SDM) has been widely validated for monofacial PV modules, its accuracy in the context of bifacial modules remains underexplored. This paper presents an advanced SDM, originally tailored for mono facial solar cells, adapted to accommodate the unique characteristics of bifacial PV modules, including the incorporation of partial shading effects. The paper explores bifacial solar cell technologies, bifacial gain parameters, and the lEC TS 60904-1-2 specification, emphasizing the significance of accurate PV modeling in system simulation. The model's accuracy is validated using manufacturer datasheets, which includes variations in $Isc$ (short-circuit current), $Voc$ (open-circuit voltage), and Pmax (maximum power output) for different bifacial gains. Additionally, a modeling approach for partial shading is presented and validated using available experimental data from previous research.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Efficacy of ADDIE Model in Peer-to-Peer Networks: Digital Evidence Investigation 8

Efficacy of ADDIE Model in Peer-to-Peer Networks: Digital Ev...

引用

8th International Conference on Future Internet of Things and Cloud, FiCloud 2021

作者： Musa, Ahmad Sanda Awan, Irfan-Ullah Abobaker, Ibrahim School of Electircal Engineering and Computer Science University of Bradford Bradford United Kingdom

ISBN: (纸本)9781665425742

While the need for content distribution proliferates-becoming more mammoth and complex on the Internet-the P2P network perseveres as one of the best avenues to service the demand for content distribution. It enjoys a wide range of clients that transport data in bits securely, making it susceptible to moving dubious contents, hence becoming exposed to varying security threats that require credible digital investigation to address. The tools and techniques used in performing digital investigations are still mostly lagging, successfully slowing down law enforcement agencies in general. The acquisition of digital evidence over the Internet is still elusive in the battle against cybercrime. This paper considers a new technique for detecting passive peers that participate in a P2P network. As part of our study, we crawled the μTorrent P2P client over 10 days while logging all participating peers. We then employed digital forensic techniques to analyze the popular users and generate evidence within them with high accuracy. Finally, we evaluated our proposed approach against the standard Analysis, Design, Development, Implementation, and Evaluation, or ADDIE model for digital investigation to arrive at the credible digital evidence presented in this paper. © 2021 IEEE.

关键词： Peer to peer networks

来源：评论

学校读者我要写书评

暂无评论

A Convex Optimization Approach to Compute Trapping Regions for Lossless Quadratic Systems

arXiv

引用

arXiv 2024年

作者： Liao, Shih-Chi Heide, A. Leonid Hemati, Maziar S. Seiler, Peter J. Electircal and Computer Engineering University of Michigan Ann ArborMI United States Aerospace Engineering and Mechanics University of Minnesota Minnesota United States

Quadratic systems with lossless quadratic terms arise in many applications, including models of atmosphere and incompressible fluid flows. Such systems have a trapping region if all trajectories eventually converge to and stay within a bounded set. Conditions for the existence and characterization of trapping regions have been established in prior works for boundedness analysis. However, prior solutions have used non-convex optimization methods, resulting in conservative estimates. In this paper, we build on this prior work and provide a convex semidefinite programming condition for the existence of a trapping region. The condition allows precise verification or falsification of the existence of a trapping region. If a trapping region exists, then we provide a second semidefinite program to compute the least conservative trapping region in the form of a ball. Two low-dimensional systems are provided as examples to illustrate the results. A third high-dimensional example is also included to demonstrate that the computation required for the analysis can be scaled to systems of up to ∼ O(100) states. The proposed method provides a precise and computationally efficient numerical approach for computing trapping regions. We anticipate this work will benefit future studies on modeling and control of lossless quadratic dynamical systems. Copyright © 2024, The Authors. All rights reserved.

关键词： Dynamical systems

来源：评论

学校读者我要写书评

暂无评论

Minimax pareto fairness: A multi objective perspective 37

Minimax pareto fairness: A multi objective perspective

引用

37th International Conference on Machine Learning, ICML 2020

作者： Martinez, Natalia Bertran, Martin Sapiro, Guillermo Department of Electircal and Computer Engineering Duke University

ISBN: (纸本)9781713821120

In this work we formulate and formally characterize group fairness as a multi-objective optimization problem, where each sensitive group risk is a separate objective. We propose a fairness criterion where a classifier achieves minimax risk and is Pareto-efficient w.r.t. all groups, avoiding unnecessary harm, and can lead to the best zero-gap model if policy dictates so. We provide a simple optimization algorithm compatible with deep neural networks to satisfy these constraints. Since our method does not require test-Time access to sensitive attributes, it can be applied to reduce worst-case classification errors between outcomes in unbalanced classification problems. We test the proposed methodology on real case-studies of predicting income, ICU patient mortality, skin lesions classification, and assessing credit risk, demonstrating how our framework compares favorably to other approaches. © 2020 37th International Conference on Machine Learning, ICML 2020. All rights reserved.

关键词： Deep neural networks

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：