检索结果-内蒙古大学图书馆

A Physical-Based Artificial Neural Networks Compact Modeling Framework for Emerging FETs

IEEE Transactions on Electron Devices 2024年第1期71卷 223-230页

作者： Yang, Ya-Shu Li, Yiming Kola, Sekhar Reddy Reddy National Yang Ming Chiao Tung University Parallel and Scientific Computing Laboratory The Institute of Communications Engineering Hsinchu300093 Taiwan National Yang Ming Chiao Tung University Parallel and Scientific Computing Laboratory Institute of Communications Engineering Institute of Biomedical Engineering Department of Electronics and Electrical Engineering Hsinchu300093 Taiwan National Yang Ming Chiao Tung University Parallel and Scientific Computing Laboratory Electrical Engineering and Computer Science International Graduate Program Hsinchu300093 Taiwan

We report a compact modeling framework based on the Grove-Frohman (GF) model and artificial neural networks (ANNs) for emerging gate-all-around (GAA) MOSFETs. The framework consists of two ANNs;the first ANN constructed with the drain current model not only can capture the main trend of device I - V characteristics but also can predict its variation even when the amount of training data for the ANN is insufficient or outside the range of applied biases. The second one is then designed to improve the model accuracy by further minimizing the errors between the target and the model outputs. We implement the proposed framework to accurately model emerging GAA nanosheet (NS) MOSFETs and complementary FETs (CFETs) without suffering from divergent issues in circuit simulation. In addition, nonphysical behaviors, such as nonzero current at zero bias, do not occur in the modeling framework. Compared to recently reported machine-learning (ML) models, our approach can achieve a similar level of model accuracy with merely 20% amount of the training data. © 1963-2012 IEEE.

关键词： Neural networks

来源：评论

学校读者我要写书评

暂无评论

Optimizing Fine-Tuning in Quantized Language Models:An In-Depth Analysis of Key Variables

引用

Computers, Materials & Continua 2025年第1期82卷 307-325页

作者： Ao Shen Zhiquan Lai Dongsheng Li Xiaoyu Hu National Key Laboratory of Parallel and Distributed Computing National University of Defense TechnologyChangsha410073China Strategic Assessments and Consultation Institute Academy of Military ScienceBeijing100091China

Large-scale Language Models(LLMs)have achieved significant breakthroughs in Natural Language Processing(NLP),driven by the pre-training and fine-tuning *** this approach allows models to specialize in specific tasks with reduced training costs,the substantial memory requirements during fine-tuning present a barrier to broader ***-Efficient Fine-Tuning(PEFT)techniques,such as Low-Rank Adaptation(LoRA),and parameter quantization methods have emerged as solutions to address these challenges by optimizing memory usage and computational *** these,QLoRA,which combines PEFT and quantization,has demonstrated notable success in reducing memory footprints during fine-tuning,prompting the development of various QLoRA *** these advancements,the quantitative impact of key variables on the fine-tuning performance of quantized LLMs remains *** study presents a comprehensive analysis of these key variables,focusing on their influence across different layer types and depths within LLM *** investigation uncovers several critical findings:(1)Larger layers,such as MLP layers,can maintain performance despite reductions in adapter rank,while smaller layers,like self-attention layers,aremore sensitive to such changes;(2)The effectiveness of balancing factors depends more on specific values rather than layer type or depth;(3)In quantization-aware fine-tuning,larger layers can effectively utilize smaller adapters,whereas smaller layers struggle to do *** insights suggest that layer type is a more significant determinant of fine-tuning success than layer depth when optimizing quantized ***,for the same discount of trainable parameters,reducing the trainable parameters in a larger layer is more effective in preserving fine-tuning accuracy than in a smaller *** study provides valuable guidance for more efficient fine-tuning strategies and opens avenues for further research into optimizing LLM

关键词： Large-scale Language Model Parameter-Efficient Fine-Tuning parameter quantization key variable trainable parameters experimental analysis

来源：评论

学校读者我要写书评

暂无评论

Compact Model Build Upon Dynamic Weighting Artificial Neural Network Approach for Complementary Field Effect Transistors

引用

IEEE Transactions on Electron Devices 2024年第1期71卷 246-253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 Institute of Pioneer Semiconductor Innovation The Institute of Artificial Intelligence Innovation National Yang Ming Chiao Tung University Parallel and Scientific Computing Laboratory Electrical Engineering and Computer Science International Graduate Program The Institute of Communications Engineering the Institute of Biomedical Engineering Department of Electronics and Electrical Engineering Hsinchu300093 Taiwan

In this work, a dynamic weighting-artificial neural network (DW-ANN) methodology is presented for quick and automated compact model (CM) generation. It takes advantage of both TCAD simulations for high accuracy and SPICE simulations for cost-effectiveness. This methodology is developed for gate-all-around (GAA) silicon (Si) nanosheet (NS) complementary field effect transistor (CFET), a potential candidate for future CMOS technology due to its innate properties. The critical process variation (PV) sources that severely degrade the CFET performance are estimated using DW-ANN. It reduces the computation cost and predicts the effects of PV sources with less than 1% error. Furthermore, a compact DW-ANN-based Verilog-A model has been developed that captures the dc as well as transient behavior accurately for circuit-level analysis. CFET-based circuits such as inverter, 6T-static random access memory (SRAM), and ring oscillator (RO) are characterized and implemented using DW-ANN-based CM. The overall average error of the model is reported as less than 2%. Therefore, the proposed device and circuit modeling approach provides a feasible solution for the rapid compact modeling of new emerging devices with good convergence and high accuracy. © 1963-2012 IEEE.

关键词： Neural networks

来源：评论

学校读者我要写书评

暂无评论

Statistical Device Simulation and Machine Learning of Process Variation Effects of Vertically Stacked Gate-All-Around Si Nanosheet CFETs

引用

IEEE Transactions on Nanotechnology 2024年 23卷 386-392页

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

In this study, we report the process variation effect (PVE) including the work function fluctuation (WKF) on the DC/AC characteristic fluctuation of stacked gate-all-around silicon complementary field-effect transistors (CFETs). The PVE affects characteristic fluctuation significantly;in particular, for the variability of off-state current. Owing to the bottom channel of a fin-type, the P-FET suffers from the worst off-state current fluctuation (more than 200% variation) compared to the N-FET. The device variability induced by the WKF is marginal because of amorphous-type metal grains. As input features to an artificial neural network (ANN) model, low and high work function values, as well as parameters of PVE that have prevalent effects on CEFT transfer characteristics are further considered and modeled. The estimated values of R2-score prove that the ANN model properly grasps information from the dataset successfully;thus, it can be used to model emerging CFETs for circuit simulation. © 2002-2012 IEEE.

关键词： Neural networks

来源：评论

学校读者我要写书评

暂无评论

A New Pipeline for Knowledge Graph Reasoning Enhanced by Large Language Models Without Fine-Tuning

A New Pipeline for Knowledge Graph Reasoning Enhanced by Lar...

引用

2024 Conference on Empirical Methods in Natural Language Processing, EMNLP 2024

作者： Chen, Zhongwu Bai, Long Li, Zixuan Huang, Zhen Jin, Xiaolong Dou, Yong National Key Laboratory of Parallel and Distributed Computing National University of Defense Technology China CAS Key Laboratory of Network Data Science and Technology Institute of Computing Technology Chinese Academy of Sciences China

ISBN: (纸本)9798891761643

Conventional Knowledge Graph Reasoning (KGR) models learn the embeddings of KG components over the structure of KGs, but their performances are limited when the KGs are severely incomplete. Recent LLM-enhanced KGR models input KG structural information into LLMs. However, they require fine-tuning on open-source LLMs and are not applicable to closed-source LLMs. Therefore, in this paper, to leverage the knowledge in LLMs without fine-tuning to assist and enhance conventional KGR models, we propose a new three-stage pipeline, including knowledge alignment, KG reasoning and entity reranking. Specifically, in the alignment stage, we propose three strategies to align the knowledge in LLMs to the KG schema by explicitly associating unconnected nodes with semantic relations. Based on the enriched KGs, we train structure-aware KGR models to integrate aligned knowledge to original knowledge existing in KGs. In the reranking stage, after obtaining the results of KGR models, we rerank the top-scored entities with LLMs to recall correct answers further. Experiments show our pipeline can enhance the KGR performance in both incomplete and general situations. © 2024 Association for Computational Linguistics.

关键词： Semantics

来源：评论

学校读者我要写书评

暂无评论

Nanosized-Metal-Grain-Pattern-Dependent Threshold-Voltage Models for the Vertically Stacked Multichannel Gate-All-Around Si Nanosheet MOSFETs and Their Applications in Circuit Simulation

引用

IEEE Transactions on Electron Devices 2024年第1期71卷 350-358页

作者： Sung, Wen-Li Li, Yiming Chuang, Min-Hui National Yang Ming Chiao Tung University Parallel and Scientific Computing Laboratory Institute of Communications Engineering Department of Electronics and Electrical Engineering Hsinchu300093 Taiwan Institute of Biomedical Engineering Institute of Pioneer Semiconductor Innovation Institute of Artificial Intelligence Innovation National Yang Ming Chiao Tung University Hsinchu300093 Taiwan

This article proposes grain-pattern-dependent threshold-voltage ( Vth ) models to predict the variability of Vth ( Vth ) due to work-function (WK) fluctuation (WKF) for the vertically stacked multichannel gate-all-around (GAA) silicon (Si) nanosheet (NS) MOSFETs (NS-FETs). In addition, the models were applied to estimate the variability of static noise margin (SNM) of a 6-T SRAM, a CMOS inverter, and a single-stage common-source amplifier. To model this phenomenon, each perturbed local metal grain is counted by the superposition principle statistically. The model can be used to explain the values of Vth by the location effect of metal grain for the vertically stacked multichannel devices. In addition, the model can predict the values of Vth for the vertically stacked multichannel devices with different metal grain sizes. Compared with the results of 3-D device simulation (3D-DS), the error rate (ER) of our model prediction is less than 0.5%. According to the formulated model, an empirical expression is further advanced, which has continuous derivatives and can be easily incorporated into a circuit simulator to assess the variability of SNM of a 6-T SRAM, CMOS inverter, and single-stage common-source amplifier affected by the WKF, where the ERs are below 0.5%. The proposed models provide a valuable approach for estimating the impact of circuit design by the WKF. © 1963-2012 IEEE.

关键词： III-V semiconductors

来源：评论

学校读者我要写书评

暂无评论

A prompt-based approach to adversarial example generation and robustness enhancement

引用

Frontiers of Computer Science 2024年第4期18卷 85-96页

作者： Yuting YANG Pei HUANG Juan CAO Jintao LI Yun LIN Feifei MA Key Lab of Intelligent Information Processing of Chinese Academy of Sciences(CAS) Institute of Computing TechnologyCASBeijing 100190China School of Computer Science and Technology University of Chinese Academy of SciencesBeijing 100049China Department of Computer Science Stanford UniversityCA 94305USA School of Computing National University of SingaporeSingapore 119077Singapore Laboratory of Parallel Software and Computational Science Institute of SoftwareChinese Academy of SciencesBeijing 100190China

Recent years have seen the wide application of natural language processing(NLP)models in crucial areas such as finance,medical treatment,and news media,raising concerns about the model robustness and *** find that prompt paradigm can probe special robust defects of pre-trained language *** prompt texts are first constructed for inputs and a pre-trained language model can generate adversarial examples for victim models via *** results show that prompt paradigm can efficiently generate more diverse adversarial examples besides synonym ***,we propose a novel robust training approach based on prompt paradigm which incorporates prompt texts as the alternatives to adversarial examples and enhances robustness under a lightweight minimax-style optimization *** on three real-world tasks and two deep neural models show that our approach can significantly improve the robustness of models to resist adversarial attacks.

关键词： robustness adversarial example prompt learning pre-trained language model

来源：评论

学校读者我要写书评

暂无评论

KC-GenRe: A Knowledge-constrained Generative Re-ranking Method Based on Large Language Models for Knowledge Graph Completion 30

KC-GenRe: A Knowledge-constrained Generative Re-ranking Meth...

引用

Joint 30th International Conference on Computational Linguistics and 14th International Conference on Language Resources and Evaluation, LREC-COLING 2024

作者： Wang, Yilin Hu, Minghao Huang, Zhen Li, Dongsheng Yang, Dong Lu, Xicheng Defense Innovation Institute Academy of Military Sciences China Information Research Center of Military Science China National Key Laboratory of Parallel and Distributed Computing China

ISBN: (纸本)9782493814104

The goal of knowledge graph completion (KGC) is to predict missing facts among entities. Previous methods for KGC re-ranking are mostly built on non-generative language models to obtain the probability of each candidate. Recently, generative large language models (LLMs) have shown outstanding performance on several tasks such as information extraction and dialog systems. Leveraging them for KGC re-ranking is beneficial for leveraging the extensive pre-trained knowledge and powerful generative capabilities. However, it may encounter new problems when accomplishing the task, namely mismatch, misordering and omission. To this end, we introduce KC-GenRe, a knowledge-constrained generative re-ranking method based on LLMs for KGC. To overcome the mismatch issue, we formulate the KGC re-ranking task as a candidate identifier sorting generation problem implemented by generative LLMs. To tackle the misordering issue, we develop a knowledge-guided interactive training method that enhances the identification and ranking of candidates. To address the omission issue, we design a knowledge-augmented constrained inference method that enables contextual prompting and controlled generation, so as to obtain valid rankings. Experimental results show that KG-GenRe achieves state-of-the-art performance on four datasets, with gains of up to 6.7% and 7.7% in the MRR and Hits@1 metric compared to previous methods, and 9.0% and 11.1% compared to that without re-ranking. Extensive analysis demonstrates the effectiveness of components in KG-GenRe. © 2024 ELRA Language Resource Association: CC BY-NC 4.0.

关键词： Knowledge graph

来源：评论

学校读者我要写书评

暂无评论

Large-Scale Deep Learning Medical Image Methodology and Applications Using Multiple GPUs 7

Large-Scale Deep Learning Medical Image Methodology and Appl...

引用

7th International Conference on Advances in Biomedical Engineering, ICABME 2023

作者： Krishnasamy, Ezhilmathi Boudisa, Selma Kanli, Georgia Marzal-Abarca, Francesc Xavier Keunen, Oliver Bouvry, Pascal University of Luxembourg Parallel Computing and Optimization Group Fstm Belval Luxembourg Luxembourg Institute of Health Translational Radiomics Belval Luxembourg

ISBN: (纸本)9798350325850

Digitalization of medical images has become firmly established in research labs and clinical settings worldwide, and its numerous benefits have helped drive its rapid expansion, which is predicted to continue momentum over the coming years. Its advantages of cutting costs, improving analysis, productivity and patient outcomes, reducing errors, and enhancing imaging and innovation will likely propel its widespread adoption. In recent years digitalization of medical images especially relied on high performance computing. This work illustrates how a large volume of digital radiology data can be analysed in a parallel heterogeneous architecture. We have used TensorFlow (including Keras) and the Horovod API to speed up the learning of Deep Learning models, particularly using GPUs in single and multiple compute nodes. We conclude that using Horovod API with data parallelism shows tremendous speed gains compared to traditional CPU when training the 2D UNet model. © 2023 IEEE.

关键词： Graphics processing unit

来源：评论

学校读者我要写书评

暂无评论

Synthetic Aperture Radar Algorithms on Transport Triggered Architecture Processors using OpenCL

Synthetic Aperture Radar Algorithms on Transport Triggered A...

引用

2023 IEEE International Radar Conference, RADAR 2023

作者： Rother, Niklas Matzner, Leonard Jaaskelainen, Pekka Leppanen, Topi Schleusner, Jens Blume, Holger Leibniz Universität Hannover Institute of Microeletronic Systems Hannover Germany Tampere University Customized Parallel Computing Group Tampere Finland

ISBN: (纸本)9781665482783

Live SAR imaging from small UAVs is an emerging field. On-board processing of the radar data requires high-performance and energy-efficient platforms. One candidate for this are Transport Triggered Architecture (TTA) processors. We implement Backprojection and Backprojection Autofocus on a TTA processor specially adapted for this task using OpenCL. The resulting implementation is compared to other platforms in terms of energy efficiency. We find that the TTA is on-par with embedded GPUs and surpasses other OpenCL-based platforms. It is outperformed only by a dedicated FPGA implementation. © 2023 IEEE.

关键词： Energy efficiency

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：