检索结果-内蒙古大学图书馆

On the Limits of Lossy Compression for Human Activity Recognition in Sensor networks

学校读者我要写书评

暂无评论

On the Limits of Lossy Compression for Human Activity Recogn...

Conference on Local computer networks (LCN)

作者： Jonas Schulz Hristina Radak Phuong T. Nguyen Giang T. Nguyen Frank H. P. Fitzek Deutsche Telekom Chair of Communication Networks TU Dresden Germany Department of Information Engineering Computer Science and Mathematics University of L’Aquila Italy Haptic Communication Systems TU Dresden Germany Centre for Tactile Internet with Human-in-the-Loop (CeTI)

Human activity recognition is crucial for tactile internet, virtual reality, and digital-twin applications. Previous works have analyzed machine learning for this purpose but often need to pay more attention to typical challenges when deploying these machine learning models in production. First, data must be transmitted to the network node on which the machine-learning model is running. However, scaling human activity recognition by the number of devices and users puts additional constraints on the available transmission channel. While transmitting less data saves bandwidth, removing redundant information from the data can also benefit machine learning. This paper addresses the problem of transmitting wearable sensor data for human activity recognition. To this end, we analyze the extent to which wearable sensor data can be compressed via sparse coding without sacrificing loss in recognition performance. We empirically illustrate, on various datasets, that only a fraction of sensor information is relevant for human activity recognition.

关键词：

Synesthesia of Machines (SoM)-Enhanced ISAC Precoding for Vehicular networks with Double Dynamics

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Yang, Zonghui Gao, Shijian Cheng, Xiang Yang, Liuqing State Key Laboratory of Advanced Optical Communication Systems and Networks School of Electronics Peking University Beijing100871 China Guangzhou511400 China Guangzhou China Department of Electronic and Computer Engineering Hong Kong University of Science and Technology Hong Kong

Integrated sensing and communication (ISAC) technology is vital for vehicular networks, yet the time-varying communication channels and rapid movement of targets present significant challenges for real-time precoding design. Traditional optimization-based methods are computationally complex and depend on perfect prior information, which is often unavailable in double-dynamic scenarios. In this paper, we propose a synesthesia of machine (SoM)-enhanced precoding paradigm that leverages modalities such as positioning and channel information to adapt to these dynamics. Utilizing a deep reinforcement learning (DRL) framework, our approach pushes ISAC performance boundaries. We also introduce a parameter-shared actor-critic architecture to accelerate training in complex state and action spaces. Extensive experiments validate the superiority of our method over existing approaches. Copyright © 2024, The Authors. All rights reserved.

关键词： Deep reinforcement learning

A Conversational Human-computer Interface for Smart Energy System Simulation Environments

学校读者我要写书评

暂无评论

A Conversational Human-Computer Interface for Smart Energy S...

Simulation Winter Conference

作者： Gabriel Dengler Peter Bazan Reinhard German Pooia Lalbakhsh Ariel Liebmann Laboratory of Computer Networks and Communication Systems Friedrich-Alexander-Universität Erlangen-Nürnberg Erlangen BY Germany Department of Data Science and AI Faculty of Information Technology Monash University Clayton VIC Australia

This paper introduces a conversational framework that enhances the usability of smart energy system simulations. This study is centered around OpenAI’s Generative Pre-trained Transformer (GPT), a fine-tuned conversational model that allows users to communicate with the system in a natural way. Therefore, users can describe their simulation scenarios in plain language and GPT seamlessly translates these descriptions into Python scripts, used as inputs to the simulation environment, in our case, AnyLogic Simulation Software. Our framework is based on the i7-AnyEnergy core framework to compute distribution flows and relevant statistics. The proposed human-machine interface facilitates and accelerates simulation modeling, as demonstrated through the two scenarios we have provided in this paper. Overall, our conversational framework has the potential to significantly improve the user experience of smart energy system simulation environments. By simplifying the interaction between users and complex simulation models, we enable users to obtain valuable insights rapidly and more easily.

关键词：

Optical coherent dot-product chip for sophisticated deep learning regression

学校读者我要写书评

暂无评论

Light(science & Applications) 2021年第12期10卷 2330-2341页

作者： Shaofu Xu Jing Wang Haowen Shu Zhike Zhang Sicheng Yi Bowen Bai Xingjun Wang Jianguo Liu Weiwen Zou State Key Laboratory of Advanced Optical Communication Systems and Networks Intelligent Microwave Lightwave Integration Innovation Center(imLic)Department of Electronic EngineeringShanghai Jiao Tong University800 Dongchuan RoadShanghai 200240China State Key Laboratory of Advanced Optical Communications System and Networks Department of ElectronicsSchool of Electronics Engineering and Computer SciencePeking UniversityBeijing 100871China Institution of Semiconductors Chinese Academy of SciencesBeijing 100083China

Optical implementations of neural networks(ONNs)herald the next-generation high-speed and energy-efficient deep learning computing by harnessing the technical advantages of large bandwidth and high parallelism of ***,due to the problems of the incomplete numerical domain,limited hardware scale,or inadequate numerical accuracy,the majority of existing ONNs were studied for basic classification *** that regression is a fundamental form of deep learning and accounts for a large part of current artificial intelligence applications,it is necessary to master deep learning regression for further development and deployment of ***,we demonstrate a silicon-based optical coherent dot-product chip(OCDC)capable of completing deep learning regression *** OCDC adopts optical fields to carry out operations in the complete real-value domain instead of in only the positive *** reusing,a single chip conducts matrix multiplications and convolutions in neural networks of any ***,hardware deviations are compensated via in-situ backpropagation control provided the simplicity of chip ***,the OCDC meets the requirements for sophisticated regression tasks and we successfully demonstrate a representative neural network,the AUTOMAP(a cutting-edge neural network model for image reconstruction).The quality of reconstructed images by the OCDC and a 32-bit digital computer is *** the best of our knowledge,there is no precedent of performing such state-of-the-art regression tasks on ONN *** is anticipated that the OCDC can promote the novel accomplishment of ONNs in modern AI applications including autonomous driving,natural language processing,and scientific study.

关键词： neural hardware coherent

LLM4WM: Adapting LLM for Wireless Multi-Tasking

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Liu, Xuanyu Gao, Shijian Liu, Boxun Cheng, Xiang Yang, Liuqing State Key Laboratory of Advanced Optical Communication Systems and Networks School of Electronics Peking University Beijing100871 China Guangzhou511400 China Guangzhou511400 China Department of Electronic and Computer Engineering The Department of Civil and Environmental Engineering The Hong Kong University of Science and Technology Hong Kong

The wireless channel is fundamental to communication, encompassing numerous tasks collectively referred to as channel-associated tasks. These tasks can leverage joint learning based on channel characteristics to share representations and enhance system design. To capitalize on this advantage, LLM4WM is proposed—a large language model (LLM) multi-task fine-tuning framework specifically tailored for channel-associated tasks. This framework utilizes a Mixture of Experts with Low-Rank Adaptation (MoE-LoRA) approach for multi-task fine-tuning, enabling the transfer of the pre-trained LLM’s general knowledge to these tasks. Given the unique characteristics of wireless channel data, preprocessing modules, adapter modules, and multi-task output layers are designed to align the channel data with the LLM’s semantic feature space. Experiments on a channel-associated multi-task dataset demonstrate that LLM4WM outperforms existing methodologies in both full-sample and few-shot evaluations, owing to its robust multi-task joint modeling and transfer learning capabilities. Copyright © 2025, The Authors. All rights reserved.

关键词： Semantics

Hybrid Near-Far Field Channel Estimation for Holographic MIMO communications

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Yue, Shaohua Zeng, Shuhao Liu, Liang Eldar, Yonina C. Di, Boya The State Key Laboratory of Advanced Optical Communication Systems and Networks School of Electronics Peking University Beijing100871 China The Department of Electronic and Information Engineering The Hong Kong Polytechnic University Hong Kong The Faculty of Mathematics and Computer Science Weizmann Institute of Science Rehovot7610001 Israel

Holographic MIMO communications, enabled by large-scale antenna arrays with quasi-continuous apertures, is a potential technology for spectrum efficiency improvement. However, the increased antenna aperture size extends the range of the Fresnel region, leading to a hybrid near-far field communication mode. The users and scatterers randomly lie in near-field and far-field zones, and thus, conventional far-field-only and near-field-only channel estimation methods may not work. To tackle this challenge, we demonstrate the existence of the power diffusion (PD) effect, which leads to a mismatch between the hybrid-field channel and existing channel estimation methods. Specifically, in far-field and near-field transform domains, the power gain of one channel path may diffuse to other positions, thus generating fake paths. This renders the conventional techniques unable to detect those real paths. We propose a PD-aware orthogonal matching pursuit algorithm to eliminate the influence of the PD effect by identifying the PD range, within which paths diffuse to other positions. PD-OMP fits a general case without prior knowledge of near-field and far-field path numbers and the user’s location. The computational complexity of PD-OMP and the Cramér-Rao Lower Bound for the sparse-signal-recovery-based channel estimation are also derived. Simulation results show that PD-OMP outperforms state-of-the-art hybrid-field channel estimation methods. © 2024, CC BY-NC-ND.

关键词： Channel estimation

Increasing Depth of Neural networks for Life-long Learning

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Kozal, Jedrzej Wozniak, Michal Department of Systems and Computer Networks Faculty of Information and Communication Technology Wroclaw University of Science and Technology Wybrzeze Wyspianskiego 27 Wroclaw50-370 Poland

Purpose: We propose a novel method for continual learning based on the increasing depth of neural networks. This work explores whether extending neural network depth may be beneficial in a life-long learning setting. Methods: We propose a novel approach based on adding new layers on top of existing ones to enable the forward transfer of knowledge and adapting previously learned representations. We employ a method of determining the most similar tasks for selecting the best location in our network to add new nodes with trainable parameters. This approach allows for creating a treelike model, where each node is a set of neural network parameters dedicated to a specific task. The Progressive Neural Network concept inspires the proposed method. Therefore, it benefits from dynamic changes in network structure. However, Progressive Neural Network allocates a lot of memory for the whole network structure during the learning process. The proposed method alleviates this by adding only part of a network for a new task and utilizing a subset of previously trained weights. At the same time, we may retain the benefit of PNN, such as no forgetting guaranteed by design, without needing a memory buffer. Results: Experiments on Split CIFAR and Split Tiny ImageNet show that the proposed algorithm is on par with other continual learning methods. In a more challenging setup with a single computer vision dataset as a separate task, our method outperforms Experience Replay. Conclusion: It is compatible with commonly used computer vision architectures and does not require a custom network structure. As an adaptation to changing data distribution is made by expanding the architecture, there is no need to utilize a rehearsal buffer. For this reason, our method could be used for sensitive applications where data privacy must be considered. © 2022, CC BY.

关键词： Knowledge management

Classification of Colorectal Cancer using ResNet and EfficientNet Models

学校读者我要写书评

暂无评论

Open Biomedical Engineering Journal 2024年 18卷

作者： Abhishek Ranjan, Abhishek Srivastva, Priyanshu Prabadevi, B. Rajagopal, Sivakumar Soangra, Rahul Subramaniam, Shamala K. School of Computer Science Engineering and Information Systems Vellore Institute of Technology Department of Sensor and Biomedical Technology Vellore Institute of Technology Crean College of Health and Behavioral Sciences Chapman University Department of Communication Technology and Networks Universiti Putra Malaysia

Introduction: Cancer is one of the most prevalent diseases from children to elderly adults. This will be deadly if not detected at an earlier stage of the cancerous cell formation, thereby increasing the mortality rate. One such cancer is colorectal cancer, caused due to abnormal growth in the rectum or colon. Early screening of colorectal cancer helps to identify these abnormal growth and can exterminate them before they turn into cancerous cells. Aim: Therefore, this study aims to develop a robust and efficient classification system for colorectal cancer through Convolutional Neural networks (CNNs) on histological images. Methods: Despite challenges in optimizing model architectures, the improved CNN models like ResNet34 and EfficientNet34 could enhance Colorectal Cancer classification accuracy and efficiency, aiding doctors in early detection and diagnosis, ultimately leading to better patient outcomes. Results: ResNet34 outperforms the EfficientNet34. Conclusion: The results are compared with other models in the literature, and ResNet34 outperforms all the other models. © 2024 The Author(s).

关键词： CNN Colorectal cancer Deep learning EfficientNetB4 Histology Learning rate Loss function Optimizer ResNet34 ROC curve Transforms

Hybrid Deep Learning Algorithm For MRI Brain Alzheimer’s Disease Prediction

学校读者我要写书评

暂无评论

Hybrid Deep Learning Algorithm For MRI Brain Alzheimer’s Di...

Energy, Materials and communication Engineering (ICEMCE), International Conference on

作者： D. Karthikeyan R. Nivasini Kiruba Thangam Raja D. Madhivadhani B. Sivakumar A.T.R. Krishna Priya Department of Electrical and Electronics Engineering SRM Institute of Science and Technology Kattankulathur Chennai Guvi Geek Networks IITM Research Park Chennai Tamilnadu India Department of Software and Systems Engineering School of Computer Science Engineering and Information Systems VIT University Vellore India Department of Electronics and Communication Engineering Kings Engineering College Sriperumbudur Chennai Department of Computer Science and Engineering Rohini College of Engineering and Technology Kanyakumari India

Magnetic resonance imaging (MRI) has been used to study the structural makeup of the brain and analyse several neurological disorders and diseased areas. For the adoption of preventative measures, early recognition of Alzheimer’s disease (AD) patients is essential. Here, a thorough inspection of the tissue arrangements obtained by MRI images of Outcome and Assessment Information (OASIS) dataset enables an exact characterization of certain brain diseases. There have been a number of division techniques for diagnosing AD that range in complexity. Compassion has been tested by deep learning techniques used to segment the structure of the brain and classify AD because they have the potential to uncover important information from vast amounts of data. In this paper the deep learning technique of Hybrid Dragonfly based GWO convolutional Neural Network (CNN) is achieved promising result for the diagnosis of AD. At image preprocessing wiener filter is used for removing the additive noises and the Gray Level Co-Occurrence Matrix (GLCM) extraction is implemented for texture analyzing. As a result, hybrid deep learning methods of CNN has the accuracy as 90% and the result of the image prediction are presented in this paper.

关键词：