检索结果-内蒙古大学图书馆

A neuromorphic core based on threshold switching memristor with asynchronous address event representation circuits

Science China(Information Sciences) 2022年第2期65卷 211-217页

作者： Jinsong WEI Jilin ZHANG Xumeng ZHANG Zuheng WU Rui WANG Jian LU Tuo SHI Mansun CHAN Qi LIU Hong CHEN School of Microelectronics University of Science and Technology of China Institute of Microelectronics of Chinese Academy of Sciences Zhejiang Lab Department of Microelectronics and Nanoelectronics Tsinghua University Department of Electronic & Computer Engineering The Hong Kong University of Science and Technology

The full memristive network hardware features high density and excellent scalability. However,recent researches on the full memristive network have been limited to a single-layer network, due to the lack of effective and flexible communication between neurons. In this design, we demonstrate a neuromorphic core based on Ag/Si O2/Au threshold switching memristor, which has built-in asynchronous address event representation(AER) circuits to provide flexible communication between neurons. Since temporally sparse spikes are the medium of communication between neurons, the AER circuits are designed to transmit spikes serially which have been encoded with neurons' addresses before transmission. With the asynchronous circuits design, the AER circuits will detect neurons' output in real-time. To test the performance of the neuromorphic core, we designed a behavioral simulator for the neuromporphic core to simulate the liquid state machine(LSM) network, which achieves a 100% recognition rate in the free spoken digital dataset. The simulation results show that the neuromorphic core obtains 35 times higher performance than the CPU and111 times higher energy efficiency than the GPU.

关键词： leaky-integration-and-fire (LIF) memristor threshold switching artificial neuron AER circuits asynchronous circuits on-chip communication

来源：评论

学校读者我要写书评

暂无评论

MNVLD: Violent Language Detection Model with Optimized Loss Function 4

MNVLD: Violent Language Detection Model with Optimized Loss ...

引用

4th International Conference on Big Data, Artificial Intelligence and Internet of Things engineering, ICBAIE 2023

作者： Li, Dongfang Liu, Jing Tan, Li Shang, Ziliang Liu, Zikang Beijing Technology and Business University School of Law Beijing100048 China Beijing Technology and Business University School of Computer Science and Engineering Beijing100048 China Chongqing Institute of Microelectronics Industry Technology Chongqing400000 China

ISBN: (纸本)9798350343618

Violent language on the Internet has become a serious problem on social platforms, posing a threat to users' sense of security and a conducive communication environment. Most existing methods focus on creating violent language dictionaries and classifying texts using support vector machines (SVM) and naive Bayesian techniques. However, these methods face challenges when confronted with tens of thousands of text comments or commenters using obscure words. Additionally, the current network violence language detection technology performs poorly in multi-domain settings. This paper aims to enhance the MFND model in the context of rumor detection. We propose the MNVLD model, which incorporates an appropriate loss function based on the characteristics of the network violence language dataset, to effectively detect violent language across multiple domains. © 2023 IEEE.

关键词： Support vector machines

来源：评论

学校读者我要写书评

暂无评论

LightMamba: Efficient Mamba Acceleration on FPGA with Quantization and Hardware Co-design

LightMamba: Efficient Mamba Acceleration on FPGA with Quanti...

引用

2025 Design, Automation and Test in Europe Conference, DATE 2025

作者： Wei, Renjie Xu, Songqiang Zhong, Linfeng Yang, Zebin Guo, Qingyu Wang, Yuan Wang, Runsheng Li, Meng Institute for Artificial Intelligence China School of Software and Microelectronics Peking University Beijing China School of Electronic and Computer Engineering Peking University Shenzhen China School of Integrated Circuits Peking University Beijing China Beijing Advanced Innovation Center for Integrated Circuits Beijing China Institute of Electronic Design Automation Peking University Wuxi China

ISBN: (纸本)9783982674100

State space models (SSMs) like Mamba have recently attracted much attention. Compared to Transformer-based large language models (LLMs), Mamba achieves linear computation complexity with the sequence length and demonstrates superior performance. However, Mamba is hard to accelerate due to the scattered activation outliers and the complex computation dependency, rendering existing LLM accelerators inefficient. In this paper, we propose LightMamba that co-designs the quantization algorithm and FPGA accelerator architecture for efficient Mamba inference. We first propose an FPGA-friendly post-training quantization algorithm that features rotation-assisted quantization and power-of-two SSM quantization to reduce the majority of computation to 4-bit. We further design an FPGA accelerator that partially unrolls the Mamba computation to balance the efficiency and hardware costs. Through computation reordering as well as fine-grained tiling and fusion, the hardware utilization and memory efficiency of the accelerator get drastically improved. We implement LightMamba on Xilinx Versal VCK190 FPGA and achieve 4.656.06 x higher energy efficiency over the GPU baseline. When evaluated on Alveo U280 FPGA, LightMamba reaches 93 tokens/s, which is 1.43 x that of the GPU baseline. © 2025 EDAA.

关键词： Integrated circuit design

来源：评论

学校读者我要写书评

暂无评论

Constructing Rectilinear Steiner Minimum Tree with Conditional Generative Adversarial Network

Constructing Rectilinear Steiner Minimum Tree with Condition...

引用

International Conference on VLSI Design

作者： Kritanta Saha Pritha Banerjee Susmita Sur-Kolay Advanced Computing and Microelectronics Unit Indian Statistical Institute Kolkata India Department of Computer Science and Engineering Sister Nivedita University Kolkata India Department of Computer Science and Engineering University of Calcutta Kolkata India

ISBN: (数字)9798331522445

ISBN: (纸本)9798331522452

Given n points in the 2D plane which are the pins of a net, a Rectilinear Steiner Minimum Tree (RSMT) which consists only of vertical and horizontal line segments is the shortest network to interconnect the pins along with a few extra (Steiner) points. An RSMT for a net has a minimum total wirelength in Manhattan distance and acts as a backbone to many state-of-the-art global routers. However, finding an RSMT is NPcomplete. In this work, we propose a method for constructing an RSMT with an image-to-image translation using a conditional generative adversarial network (cGAN) based model. Images of RSMTs with optimal wirelength for randomly generated nets form our training dataset. Experimental results show that, after training, our model generates RSMTs with $\leq 0.348\%$ wirelength error on average compared with the optimal solution for nets with degrees up to 50. This is the first successful attempt to solve the RSMT construction problem as an image-to-image translation problem using a cGAN based model.

关键词： Training Steiner trees Translation Computational modeling Machine learning Very large scale integration Generative adversarial networks Generators Pins Testing

来源：评论

学校读者我要写书评

暂无评论

Si–Ni-alloy-assisted very high-areal-capacity silicon-based anode on Ni foam for lithium ion battery

引用

Materials Today Chemistry 2023年 30卷

作者： Tzeng, Y. Jhan, C.-Y. Chiu, K.-M. Wu, Y.-C. Chen, G.-Y. Wang, P.-S. Institute of Microelectronics Department of Electrical Engineering College of Electrical and Computer Engineering National Cheng Kung University Tainan Taiwan

Silicon is investigated as an active material for the anode of lithium-ion battery (LIB) because of the 10 times higher specific capacity of silicon than graphite. In order to achieve a low internal resistance and high areal capacity of a silicon-based anode, metal foams are used as three-dimensional current collectors. However, unless the foam is densely compressed, active materials adhere loosely to the foam and suffer from loss of electrical contacts with the current collector. A densely compressed anode suffers from poor physical integrity without spare room for large volume expansion of silicon. By means of pyrolysis at 800 °C in argon of a mixture of silicon nanoparticles, phenolic resin binder, and Super P conductivity agent on the nickel foam, silicon reacts with nickel to form nickel silicide while the specially selected binder is graphitized to form a conductive and porous anode with a nickel foam current collector. We report 80-nm-silicon-based anode with a nickel foam, exhibiting excellent cycling performance and a high retained areal capacity of 6.5 mAh/cm2 after 50 cycles under a discharge/charge current density of 4 mA/cm2. For anode area of 1 cm2, it stores 23.4 C of charges at a current rate of 4 mA. © 2023 Elsevier Ltd

关键词： Anodes

来源：评论

学校读者我要写书评

暂无评论

A Comprehensive Evaluation of Fake Face Recognition Scheme using Artificial Intelligence Oriented Learning Scheme 1

A Comprehensive Evaluation of Fake Face Recognition Scheme u...

引用

1st International Conference on Pioneering Developments in computer Science and Digital Technologies, IC2SDT 2024

作者： Subashree, S. Rose S, Remya Valarmathy, A.S. Joseph, Princy Dennis, P. Jasmin Ravi, S. E.G.S. Pillay Engineering College Department of Computer Science and Engineering Tamil Nadu Nagapattinam India R.M.D. Engineering College Department of Artificial Intelligence and Machine Learning Tamil Nadu Chennai India Prince Shri Venkateshwara Padmavathy Engineering College Department of Electrical and Electronics Engineering Tamil Nadu Chennai India CMR College of Engineering and Technology Department of Computer Science and Engineering Hyderabad India Vi Institute of Technology Department of Electronics and Communication Engineering Tamil Nadu Chengalpattu India Saveetha University Department of VLSI Microelectronics Tamil Nadu Chennai India

ISBN: (纸本)9798350365016

Fake face recognition has emerged as a critical area of research due to the proliferation of synthetic media and its potential misuse in various domains. This study presents a thorough evaluation of a novel Fake Face Recognition Scheme utilizing Artificial Intelligence Oriented Learning Scheme (AIOLS) coupled with Generative Adversarial Network (GAN)-GoogleNet integration. The aim is to discern authentic facial images from synthetic or tampered ones with high accuracy. In this research, we employed a dataset comprising a diverse range of authentic and synthetic facial images. The proposed scheme incorporates the power of GANs for generating synthetic facial images and leverages GoogleNet, a state-of-the-art deep convolutional neural network, for discriminative feature extraction. The integration within the AIOLS framework facilitates efficient learning and adaptation to complex patterns present in fake facial images. The evaluation of our scheme was conducted through rigorous experimentation on the dataset, employing various performance metrics. Notably, the accuracy obtained was measured at an impressive 96.7%, signifying the effectiveness of our approach in distinguishing between real and fake facial images. Additionally, other metrics such as precision, recall, and F1-score were also computed to provide a comprehensive assessment of the scheme's performance. © 2024 IEEE.

关键词： Generative adversarial networks

来源：评论

学校读者我要写书评

暂无评论

Discrete memristive neuron model and its interspike interval-encoded application in image encryption

引用

Science China(Technological Sciences) 2021年第10期64卷 2281-2291页

作者： BAO Han HUA ZhongYun LIU WenBo BAO BoCheng College of Automation Engineering Nanjing University of Aeronautics and AstronauticsNanjing 210016China School of Computer Science and Technology Harbin Institute of TechnologyShenzhenShenzhen 518055China School of Microelectronics and Control Engineering Changzhou UniversityChangzhou 213164China

Bursting is a diverse and common phenomenon in neuronal activation patterns and it indicates that fast action voltage spiking periods are followed by resting *** interspike interval(ISI)is the time between successive action voltage spikes of neuron and it is a key indicator used to characterize the ***,a three-dimensional memristive Hindmarsh-Rose(mHR)neuron model was constructed to generate hidden chaotic ***,the properties of the discrete mHR neuron model have not been investigated,*** this article,we first construct a discrete mHR neuron model and then acquire different hidden chaotic bursting sequences under four typical sets of *** make these sequences more suitable for the application,we further encode these hidden chaotic sequences using their ISIs and the performance comparative results show that the ISI-encoded chaotic sequences have much more complex chaos properties than the original *** addition,we apply these ISI-encoded chaotic sequences to the application of image *** image encryption scheme has a symmetric key structure and contains plain-text permutation and bidirectional diffusion *** results and security analyses prove that it has excellent robustness against various possible attacks.

关键词： chaos complexity chaotic bursting sequence memristive neuron model interspike interval(ISI) image encryption

来源：评论

学校读者我要写书评

暂无评论

Web-Based AI Assistant for Medical Imaging: A Case Study on Predicting Spontaneous Preterm Birth via Ultrasound Images 25th

Web-Based AI Assistant for Medical Imaging: A Case Study on...

引用

25th International Conference on Web Information Systems engineering, WISE 2024

作者： Bi, Weichen Shao, Zijian Han, Yudong Du, Jiaqi Wei, Yuan Guo, Lijuan Wu, Tianchen Li, Shuang Ma, Yun Institute for Artificial Intelligence Peking University Beijing China School of Electronics Engineering and Computer Science Peking University Beijing China Peking University Third Hospital Beijing China National Centre for Healthcare Quality Management in Obstetrics Beijing China School of Software and Microelectronics Peking University Beijing China

ISBN: (纸本)9789819605729

The potential for artificial intelligence (AI) in analyzing medical images is vast and promises significant future advancements. It brings opportunities for community and remote-area hospitals to be equipped with professional capabilities once exclusive to top-tier medical institutions. However, applying in-lab AI methods to real-world applications of medical imaging is challenging due to the complexity of gathering training datasets as well as the need for intricate systems and specialized devices. In this paper, we demonstrate how the web platform could benefit the application of AI methods in medical imaging based on the lightweight design, cross-platform portability, streamlined distribution and deployment of the web. Specifically, we design and implement a web-based assistant for predicting spontaneous preterm births via ultrasound images. During the development phase, we leverage crowdsourcing on the web to annotate ultrasound images and gather domain-specific features to train the AI model for predicting spontaneous preterm birth. During the deployment phase, we employ WebAR to present AI-assisted diagnostic insights for physicians. Evaluation results show that our system achieves an AUC of 0.769, nearing the diagnostic proficiency of top-tier physicians. Besides, our WebAR system exhibits only 527.2–1754.2 ms latency, enabling effective assisted diagnosis. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025.

关键词： Ultrasonic imaging

来源：评论

学校读者我要写书评

暂无评论

The reliability challenge of SiC Power Modules in Automotive Applications 14

The reliability challenge of SiC Power Modules in Automotive...

引用

14th International Conference on Advanced Semiconductor Devices and Microsystems, ASDAM 2022

作者： Calabretta, Michele Sitta, Alessandro Mauromicale, Giuseppe Rundo, Francesco Sequenzia, Gaetano Messina, Angelo Alberto STMicroelectronics Catania95121 Italy University of Catania Department of Electrical Electronic and Computer Engineering Catania95125 Italy Institute for Microelectronics and Microsystems Cnr Catania95121 Italy

ISBN: (数字)9781665469791

ISBN: (纸本)9781665469777

Silicon carbide (SiC)-based power modules in automotive applications are becoming more and more important in the framework of battery and hybrid vehicles. Consequently, the reliability concerns related to these products must be carefully assessed, considering the harsh environment of automotive applications. The aim of this work is to give some insights into the reliability assessments during the design stage of SiC modules devoted to traction applications, considering different aspects such as power cycle, thermal characterization, and solder joint reliability. © 2022 IEEE.

关键词： Silicon carbide

来源：评论

学校读者我要写书评

暂无评论

Analysis of the Performance of Software Defined Networks (SDN) Versus Networks with TCP/IP Architecture 2

Analysis of the Performance of Software Defined Networks (SD...

引用

2nd International Conference on Intelligent Data Communication Technologies and Internet of Things, IDCIoT 2024

作者： Tanguturi, Rama Chaithanya Amutharaj, J. Prakash, N. Subramani, K. Taqui, Syed Noeman Sree Southry, S. Pace Institute of Technology and Sciences Department of Computer Science and Engineering Andhra Pradesh Ongole India Raja Rajeswari College of Engineering Department of Information Science and Engineering Bengaluru560074 India Kumaraguru College of Technology Department of Electrical and Electronics Engineering Coimbatore India Department of Physics Tamil Nadu Dindigul India Saveetha School of Engineering Department of VLSI Microelectronics Saveetha Institute of Medical and Technical Sciences Tamil Nadu Chennai602105 India Sona College of Technology Department of Electronics and Communication Engineering Tamilnadu Salem India

ISBN: (纸本)9798350327533

It is expected that the transfer speeds on average would rise from 39Mbps to 75.4Mbps. For this reason, the idea of implementing centralized networks, with easy administration and traffic management through policies, has become essential in recent years. Software-defined networks (SDN) and traditional TCPIIP architecture networks are two distinct approaches to network design and management. While TCPIIP architecture networks have been the norm for decades, SDN is a relatively new concept that promises to revolutionize network performance, flexibility, and scalability. In this work, the performance of SDN versus networks with TCPIIP architecture is analyzed and compared. In the analysis of the performance of software-defined networks and networks with TCPIIP architecture, it is predicted to have a favourable response in the evaluation of different metrics such as latency, jitter and bandwidth in both networks;but with better performance in the case of SDN networks. © 2024 IEEE.

关键词： Flexibility and Adaptability Integration with Cloud Services Management and Configuration Security Software-defined networks (SDN)

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：