检索结果-内蒙古大学图书馆

2021 International Conference on Electronic Information Engineering and Computer Communication, EIECC 2021

作者： Wang, Zhenpeng Cheng, Jixiang Wu, Dan He, HongBin Xiao, Shu School of Electrical Information Southwest Petroleum University Sichuan Province Chengdu610000 China

ISBN: (数字)9781510653207

ISBN: (纸本)9781510653191

The design of an efficient and compact convolutional neural network (CNN) for image classification is a very challenging problem, and its design process relies heavily on the experience of experts and continuous trial and error. This paper proposed a search method based on variable scale convolutional neural network by comprehensively considering the convolutional layer, pooling layer, fully-connection layer, and activation function in CNN. In this method, a search space is designed, and the categories of search parameters are maximized. At the same time, a particle swarm variable length coding mapping method is proposed to solve the problem of coding redundancy of candidate networks. In order to effectively evaluate the performance of candidate networks, the evaluation method of the random dataset is adopted to reduce the evaluation time of CNN and improve the stability of training. The proposed search model is compared with many empirical design models and other search methods in four image classification tasks. The experimental results show that the proposed method has strong advantages and competitiveness in classification accuracy and model size compared with the existing methods. © COPYRIGHT SPIE. Downloading of the abstract is permitted for personal use only.

关键词： Mapping

来源：评论

学校读者我要写书评

暂无评论

News Categorization based on Titles with SVM, Naïve Bayesian, Random Forest, and RNN algorithms 2

News Categorization based on Titles with SVM, Naïve Bayesia...

引用

2nd International Conference on Artificial Intelligence, Automation, and High-Performance computing, AIAHPC 2022

作者： Li, Yongwei Liu, Kejun Liu, Ziyu Tao, Zhen Yuan, Meng Department of Computer Science and Technology Tsinghua University Beijing China College of Engineering Cornell University IthacaNY United States Institute of Physics University College London London United Kingdom Electrical and Electrical Department The University of Bath Bath United Kingdom School of Economics and Management Tsinghua University Beijing China

ISBN: (纸本)9781510657717

News categorization, a text classification task, is now commonly used in many news websites. However, many of these news classifiers require full content of the news, which would cost great amounts of time for computation. In this paper, we focus on the possibility of categorizing news by its title with Support Vector Machines, Random Forest Classifiers, Naive Bayes, and Recurrent neural network. First, we explore some widely used pre-processing methods, including Bag of Words and Word2Vec. Then we combine these different pre-processing methods with the machine learning algorithms mentioned above to create different models. We measure their performances on the News Aggregator Data Set from UCI Machine Learning Repository, which contains over 400,000 pieces of news over 4 main categories. To evaluate the related performances, we use 85% data as a training set and 5% data as a validation set, and finally, use 10% data as a testing set. Comprehensive experimental results demonstrate that even with only the news titles, some models can still perform well in this challenging task. Therefore, it is possible to categorize news through its title in high accuracy yet with a much lower computing cost compared to full-text classification. © 2022 SPIE. All rights reserved.

关键词： Recurrent neural networks

来源：评论

学校读者我要写书评

暂无评论

DRB-Net: Dilated Residual Block network for Infrared Image Restoration 17th

DRB-Net: Dilated Residual Block Network for Infrared Image R...

引用

17th International Symposium on Visual computing (ISVC)

作者： Falahkheirkhah, Kianoush Yeh, Kevin Confer, Matthew P. Bhargava, Rohit Dept Chem & Biomol Engn Urbana IL 61801 USA Beckman Inst Adv Sci & Technol Urbana IL 61801 USA Univ Illinois Canc Ctr Illinois Urbana IL 61801 USA

ISBN: (纸本)9783031207150;9783031207167

Infrared (IR) spectroscopic imaging offers label-free visualization of sample heterogeneity via spatially localized chemical information. This spatial-spectral data set is amenable to computational algorithms that highlight functional properties of the sample. Although Fourier transform IR (FT-IR) imaging provides reliable analytical information over a wide spectral profile, long data acquisition times are a major challenge impeding broad adoptability. Discrete frequency (DF) IR imaging is considerably faster, first by reducing the total number of spectral frequencies acquired to only those necessary for the task, and second by using substantially higher optical power via IR lasers. Further acceleration of imaging is hindered by high laser noise and usually relies on time-consuming averaging of ensemble measurements to achieve useful signal-to-noise ratio (SNR). Here, we develop a novel convolutional neural network (CNN) architecture capable of denoising discrete frequency infrared (DFIR) images in real-time, removing the need for excessive co-averaging, thereby reducing the total data acquisition time accordingly. Our architecture is based on dilated residual block network (DRB-Net), which outperforms state-of-the-art CNN models for image denoising task. To validate the robustness of DRB-Net, we demonstrate its efficacy on various unseen samples including SU-8 targets, polymers, cells, and prostate tissues. Our findings demonstrate that DRB-Net recovers high-quality data from noisy input without supervision and with minimal computation time.

关键词： Infrared imaging Image enhancement Computer vision Denoising

来源：评论

学校读者我要写书评

暂无评论

Comparison of two deep learning methods for ship target recognition with optical remotely sensed data

引用

neural computing & APPLICATIONS 2021年第10期33卷 4639-4649页

作者： Zhang, Dianjun Zhan, Jie Tan, Lifeng Gao, Yuhang Zupan, Robert Tianjin Univ Sch Marine Sci & Technol Tianjin 300072 Peoples R China Tianjin Univ Dept Culture & Tourism Informat Technol Architect Key Lab Sch Architecture Tianjin 300072 Peoples R China Harbin Engn Univ Coll Underwater Acoust Engn Harbin 150001 Peoples R China Univ Zagreb Fac Geodesy Zagreb 10000 Croatia

As an important part of modern marine monitoring systems, ship target identification has important significance in maintaining marine rights and monitoring maritime traffic. With the development of artificial intelligence technology, image detection and recognition based on deep learning methods have become the most popular and practical method. In this paper, two deep learning algorithms, the Mask R-CNN algorithm and the Faster R-CNN algorithm, are used to build ship target feature extraction and recognition models based on deep convolutional neural networks. The established models were compared and analyzed to verify the feasibility of target detection algorithms. In this study, 5748 remote sensing maps were selected as the dataset for experiments, and two algorithms were used to classify and extract warships and civilian ships. Experiments showed that for the accuracy of ship identification, Mask R-CNN and Faster R-CNN reached 95.21% and 92.76%, respectively. These results demonstrated that the Mask R-CNN algorithm achieves pixel-level segmentation. Compared with the Faster R-CNN algorithm, the obtained target detection effect is more accurate, and the performance in target detection and classification is better, which reflects the great advantage of pixel-level recognition.

关键词： Full convolutional network Ship target recognition Pixel level Mask R-CNN Faster R-CNN optical remote sensing images

来源：评论

学校读者我要写书评

暂无评论

A System for Simultaneous Optogenetics Stimulation, Calcium Imaging, Electrical Signal Detection, and Electrical Stimulation of In Vitro Neuronal networks

A System for Simultaneous Optogenetics Stimulation, Calcium ...

引用

Asia Communications and Photonics Conference and Exhibition (ACP)

作者： Yin Deng Yarong Lin Shiyang Cao Tao Li Jie Li Zeying Lu Yueheng Lan Xiaojuan Sun Longze Sha Lili Gui Qi Xu Kun Xu State Key Laboratory of Information Photonics and Optical Communications School of Electronic Engineering Beijing University of Posts and Telecommunications Beijing China Department of Biochemistry and Molecular Biology State Key Laboratory of Common Mechanism Research for Major Diseases Chinese Academy of Medical Sciences and Peking Union Medical College Beijing China State Key Laboratory of Information Photonics and Optical Communications School of Science Beijing University of Posts and Telecommunications Beijing China

ISBN: (数字)9798350379266

ISBN: (纸本)9798350379273

In vitro neuronal networks are promising models for neural computing. Despite numerous efforts to determine the activities of these neuronal networks, achieving high spatiotemporal resolution in recording and controlling neuronal activities remains challenging. Here, we propose the OptoCaMEA system, which simultaneously records neuronal network activities with a temporal resolution of 50 microseconds and a spatial resolution smaller than 10 micrometres. Additionally, it achieves single-neuron control with a spatial resolution smaller than 10 micrometres. OptoCaMEA offers a more precise method for recording and controlling the dynamics of in vitro neuronal networks, facilitating the development of computational frameworks based on biological neural networks and contributing to the in-depth study of related neurological disease models.

关键词： Micrometers Computational modeling Optogenetics Recording Spatiotemporal phenomena Spatial resolution In vitro Biological neural networks Signal detection Photonics

来源：评论

学校读者我要写书评

暂无评论

Single-shot optical neural network

引用

SCIENCE ADVANCES 2023年第25期9卷 eadg7904页

作者： Bernstein, Liane Sludds, Alexander Panuski, Christopher Trajtenberg-Mills, Sivan Hamerly, Ryan Englund, Dirk MIT Res Lab Elect 50 Vassar St Cambridge MA 02139 USA Lightmatter Inc 100 Summer St Boston MA 02110 USA NTT Res Inc Phys & Informat Labs Sunnyvale CA 94085 USA

Analog optical and electronic hardware has emerged as a promising alternative to digital electronics to improve the efficiency of deep neural networks (DNNs). However, previous work has been limited in scalability (input vector length K approximate to 100 elements) or has required nonstandard DNN models and retraining, hindering widespread adoption. Here, we present an analog, CMOS-compatible DNN processor that uses free-space optics to reconfigurably distribute an input vector and optoelectronics for static, updatable weighting and the nonlinearity-with K approximate to 1000 and beyond. We demonstrate single-shot-per-layer classification of the MNIST, Fashion-MNIST, and QuickDraw datasets with standard fully connected DNNs, achieving respective accuracies of 95.6, 83.3, and 79.0% without preprocessing or retraining. We also experimentally determine the fundamental upper bound on throughput (similar to 0.9 exaMAC/s), set by the maximum optical bandwidth before substantial increase in error. Our combination of wide spectral and spatial bandwidths enables highly efficient computing for next-generation DNNs.

关键词： Deep neural networks

来源：评论

学校读者我要写书评

暂无评论

Enhancing Odia HCR: Tracing, Zoning and neural networks 5

Enhancing Odia HCR: Tracing, Zoning and Neural Networks

引用

5th International Conference on Emerging Systems and Intelligent computing, ESIC 2025

作者： Mishra, Soumya Ranjan Pramanik, Subhadip Pathak, Debanjan Mohapatra, Hitesh University School of Computer Engineering Odisha Bhubaneswar751024 India

ISBN: (纸本)9798331522100

Recognition of handwritten characters remains a challenging task, especially across diverse languages and scripts. optical character recognition (OCR) plays a pivotal role in transforming images of handwritten documents into machine-readable text. Whether performed online (during documentation) or offline (from stored documents), OCR is utilized in a variety of fields, including medical transcription, language translations, the digitization of ancient manuscripts, and more. In this work, we propose an efficient framework for handwritten character recognition, applicable to both offline and online processes. Our approach involves: We preprocess handwritten character images to make them suitable for feature extraction. Instead of relying solely on pixel intensities, we extract global and local shape features. These features capture the essence of character shapes. We employ a Kohonen neural network (SOM) to learn the underlying patterns in handwritten characters. Our experiments on the NITRKL dataset demonstrate the superiority of our models. By understanding the strokes, medium, and veracity of the data, we achieve high-quality recognition results. © 2025 IEEE.

关键词： Self organizing maps

来源：评论

学校读者我要写书评

暂无评论

Noise-Resistant optical Implementation of Analogue neural networks

Noise-Resistant Optical Implementation of Analogue Neural Ne...

引用

Conference on Lasers and Electro-Optics Europe / European Quantum Electronics Conference (CLEO/Europe-EQEC)

作者： Ron, Diego Arguello Kamalian-Kopae, Morteza Turitsyn, Sergei Aston Univ Aston Inst Photon Technol Birmingham B4 7ET W Midlands England

ISBN: (纸本)9781665418768

Analogue artificial neural networks are widely considered as promising computational models that more closely imitate the information processing capabilities of the human brain compared to digital neural networks. The significant computation power and the much reduced power consumption per operation make the analogue implementation of neural networks very attractive. There is an active research on artificial neural networks (ANNs) implementation using both analogue photonic and electronic hardware [1] – [4] . However, compared to digital realisations the conventional analogue systems are more sensitive to the noise that is inevitably present in practical implementations [2] , [3] . Noise properties in ANNs have been studied both in the electronic and photonic domains. However, photonic ANNs are much less investigated compared to the electronic implementations, for which some training techniques have been proposed to enhance ANNs robustness against noise [1] , [4] .

关键词： Training Power demand Artificial neural networks optical computing optical fiber networks Robustness optical sensors

来源：评论

学校读者我要写书评

暂无评论

Motion Vector Extrapolation for Video Object Detection

引用

JOURNAL OF IMAGING 2023年第7期9卷 132-132页

作者： True, Julian Khan, Naimul Toronto Metropolitan Univ Dept Elect Comp & Biomed Engn Toronto ON M5B 2K3 Canada

Despite the continued successes of computationally efficient deep neural network architectures for video object detection, performance continually arrives at the great trilemma of speed versus accuracy versus computational resources (pick two). Current attempts to exploit temporal information in video data to overcome this trilemma are bottlenecked by the state of the art in object detection models. This work presents motion vector extrapolation (MOVEX), a technique which performs video object detection through the use of off-the-shelf object detectors alongside existing optical flow-based motion estimation techniques in parallel. This work demonstrates that this approach significantly reduces the baseline latency of any given object detector without sacrificing accuracy performance. Further latency reductions up to 24 times lower than the original latency can be achieved with minimal accuracy loss. MOVEX enables low-latency video object detection on common CPU-based systems, thus allowing for high-performance video object detection beyond the domain of GPU computing.

关键词： object detection video convolutional neural network motion vector optical flow

来源：评论

学校读者我要写书评

暂无评论

4th International Conference on Advanced network Technologies and Intelligent computing, ANTIC 2024

4th International Conference on Advanced Network Technologie...

引用

4th International Conference on Advanced network Technologies and Intelligent computing, ANTIC 2024

ISBN: (纸本)9783031837951

The proceedings contain 109 papers. The special focus in this conference is on Advanced network Technologies and Intelligent computing. The topics include: Exploratory Study on Different Transformer models;visionLingua: Empowering Multimodal Understanding with Cutting-Edge Language models;human Action Recognition Using Mediapipe Holistic Keypoints: A Deep Learning Approach;machine Learning for Waste-to-Energy: Optimization and Predictive Analytics;enhanced Traffic Forecasting for Urban Planning Using DeepWalk Embeddings and Spatio-Temporal Graph Ordinary Differential Equation Model;fault Detection in Semiconductor Manufacturing Using Advanced Machine Learning Techniques with Interpretability;fine-Tuning RetinaNet for Few-Shot Fruit Detection;Smart Street: AI-Powered Traffic Flow Enhancement with Adaptive Signal Control;a Hybrid Machine Learning Approach for Accurate Milk Quality Classification;Designing VoiceGPT Chatterbox Concierge: A Personal AI Assistant for Daily Services Using Supervised Machine Learning Based Speech Recognition Techniques;improved Classification Techniques Using Convolutional neural networks for Rice Plant Disease Detection;Enhancing Customer Segmentation and Behavior Prediction Through Machine Learning and Medium Gaussian SVM;Grapevine SDLC Model for Real-Time Fake News Classification;Detecting AI-Generated Text Using Fine-Tuned Transformers: A Study on Academic Integrity;MoReco: AI Model Recommendation and Optimization for B5G networks;file Prefetching Accuracy Enhancement Using Deep Learning Approaches;Mamba Based Conditional GAN for optical Cloud Removal of Satellite Images;a Hybrid Quantum Machine Learning Model for Multi-class Classifier;adaptive Resource Management in Massive Machine-Type Communications Using Reinforcement Learning;Advanced Machine Learning and Features Based Data Analytics of Electricity Consumption in USA and India;cross-Domain Adversarial Attacks and Robust Defense Mechanisms for Multimodal neural networks;deep Featu

关键词：

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：