检索结果-内蒙古大学图书馆

An efficient deep learning model for early disease detection in vegetable crops

Neural Computing and Applications 2025年 1-20页

作者： Bhola, Amit Kumar, Prabhat School of Computer Science and Engineering Sharda University Greater Noida201310 India Department of Computer Science and Engineering National Institute of Technology Patna Bihar Patna800005 India

Plant diseases significantly threaten global food security and economic stability by reducing crop yields, increasing production costs, and exacerbating food shortages. Early and precise detection of plant diseases is essential for mitigating these risks. This study introduces a lightweight deep learning model for early and efficient disease detection in key vegetable crops, including eggplant, potato, tomato, and soybean. The model leverages a comprehensive dataset containing 23 classes of healthy and diseased plant images. With only 388,055 parameters and a model size of 1.48 MB, it achieves an impressive accuracy of 92.75%, outperforming 17 state-of-the-art deep learning models in terms of performance, efficiency, and size. The novelty lies in the model’s compact architecture, making it highly suitable for deployment in resource-constrained environments such as mobile and edge devices. The primary objective is to provide a scalable and cost-effective solution for early disease detection to enhance agricultural productivity. The findings underscore the model’s superiority over traditional methods and emphasize its potential for real-world applications in smart and sustainable agriculture systems. © The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature 2025.

关键词： Smart agriculture

来源：评论

学校读者我要写书评

暂无评论

Improving meta-learning model via meta-contrastive loss

引用

Frontiers of computer science 2022年第5期16卷 107-113页

作者： Pinzhuo TIAN Yang GAO Department of Computer Science and Technology Nanjing UniversityJiangsu 210023China

Recently,addressing the few-shot learning issue with meta-learning framework achieves great *** we know,regularization is a powerful technique and widely used to improve machine learning ***,rare research focuses on designing appropriate meta-regularizations to further improve the generalization of meta-learning models in few-shot *** this paper,we propose a novel metacontrastive loss that can be regarded as a regularization to fill this *** motivation of our method depends on the thought that the limited data in few-shot learning is just a small part of data sampled from the whole data distribution,and could lead to various bias representations of the whole data because of the different sampling ***,the models trained by a few training data(support set)and test data(query set)might misalign in the model space,making the model learned on the support set can not generalize well on the query *** proposed meta-contrastive loss is designed to align the models of support and query sets to overcome this *** performance of the meta-learning model in few-shot learning can be *** experiments demonstrate that our method can improve the performance of different gradientbased meta-learning models in various learning problems,e.g.,few-shot regression and classification.

关键词： meta-learning few-shot learning metaregularization deep learning

来源：评论

学校读者我要写书评

暂无评论

Integrating Transformer and Bidirectional Long Short-Term Memory for Intelligent Breast Cancer Detection from Histopathology Biopsy Images

引用

computer Modeling in engineering & sciences 2024年第10期141卷 443-458页

作者： Prasanalakshmi Balaji Omar Alqahtani Sangita Babu Mousmi Ajay Chaurasia Shanmugapriya Prakasam Department of Computer Science College of Computer ScienceKing Khalid UniversityAbha61451Saudi Arabia College of Science and Arts in Rijal Alma King Khalid UniversityAbha61451Saudi Arabia Department of Information Technology Muffakham Jah College of Engineering and TechnologyHyderabad500034India Department of Computer Science and Business Systems Rajalakshmi Engineering CollegeThandalam602105India

Breast cancer is a significant threat to the global population,affecting not only women but also a threat to the entire *** recent advancements in digital pathology,Eosin and hematoxylin images provide enhanced clarity in examiningmicroscopic features of breast tissues based on their staining *** cancer detection facilitates the quickening of the therapeutic process,thereby increasing survival *** analysis made by medical professionals,especially pathologists,is time-consuming and challenging,and there arises a need for automated breast cancer detection *** upcoming artificial intelligence platforms,especially deep learning models,play an important role in image diagnosis and ***,the histopathology biopsy images are taken from standard data ***,the gathered images are given as input to the Multi-Scale Dilated Vision Transformer,where the essential features are ***,the features are subjected to the Bidirectional Long Short-Term Memory(Bi-LSTM)for classifying the breast cancer *** efficacy of the model is evaluated using divergent *** compared with other methods,the proposed work reveals that it offers impressive results for detection.

关键词： Bidirectional long short-term memory breast cancer detection feature extraction histopathology biopsy images multi-scale dilated vision transformer

来源：评论

学校读者我要写书评

暂无评论

Class-conditional domain adaptation for semantic segmentation

引用

Computational Visual Media 2024年第5期10卷 1013-1030页

作者： Yue Wang Yuke Li James H.Elder Runmin Wu Huchuan Lu School of Information and Communication Engineering Dalian University of TechnologyDalian 116024China School of Computer Science Wuhan UniversityWuhan 430072China Department of Electrical Engineering and Computer Science York UniversityToronto M3J 1P3Canada Department of Computer Science the University of Hong KongHong Kong 999077China

Semantic segmentation is an important sub-task for many ***,pixel-level ground-truth labeling is costly,and there is a tendency to overfit to training data,thereby limiting the generalization *** domain adaptation can potentially address these problems by allowing systems trained on labelled datasets from the source domain(including less expensive synthetic domain)to be adapted to a novel target *** conventional approach involves automatic extraction and alignment of the representations of source and target domains *** limitation of this approach is that it tends to neglect the differences between classes:representations of certain classes can be more easily extracted and aligned between the source and target domains than others,limiting the adaptation over all ***,we address:this problem by introducing a Class-Conditional Domain Adaptation(CCDA)*** incorporates a class-conditional multi-scale discriminator and class-conditional losses for both segmentation and ***,they measure the segmentation,shift the domain in a classconditional manner,and equalize the loss over *** results demonstrate that the performance of our CCDA method matches,and in some cases,surpasses that of state-of-the-art methods.

关键词： domain adaptation generative adversarial networks semantic segmentation cityscapes

来源：评论

学校读者我要写书评

暂无评论

A Novel Enhanced Approach for Security and Privacy Preserving in IoT Devices with Federal Learning Technique

引用

SN computer science 2024年第6期5卷 750页

作者： Moeed, Syed Abdul Karnati, Ramesh Ashmitha, G. Mohammad, Gouse Baig Mohanty, Sachi Nandan Department of Computer Science and Engineering Kakatiya Institute of Technology and Science Department of Computer Science and Engineering Vardhaman College of Engineering School of Computer Science & amp Engineering (SCOPE) VIT-AP University

The Internet of Things (IoT) has revolutionized our lives, but it has also introduced significant security and privacy challenges. The vast amount of data collected by these devices, often containing sensitive information, makes them prime targets for cyberattacks. Traditional security methods struggle to keep pace with the evolving threat landscape of the interconnected IoT world. In response, this paper explores the transformative potential of federated learning (FL) in safeguarding both the privacy and security of IoT devices. FL keeps data on individual devices, only sharing updated models, not raw data, thus protecting user privacy and eliminating the need for a central data storage server, which reduces the risk of data breaches and security vulnerabilities. The effectiveness of the proposed model is evaluated using established open-source datasets like NSL-KDD and UNSW NB15, ensuring real-world applicability. Analysis of the dataset's features enabled the development of a model utilizing federated learning. Notably, the proposed model achieved superior performance with FL in detecting attacks on IoT networks. Furthermore, this research investigates the transformative potential of FL to address the inherent security and privacy challenges plaguing traditional, centralized data collection methods in the ever-expanding realm of IoT devices. FL empowers collaborative learning on distributed datasets, allowing devices to collectively improve security without compromising user privacy. The findings of this study demonstrate the promise of FL as a secure and privacy-preserving solution for the future of IoT. © The Author(s), under exclusive licence to Springer Nature Singapore Pte Ltd. 2024.

关键词： Cyberattacks Federal learning Internet of Things Machine learning Security and privacy

来源：评论

学校读者我要写书评

暂无评论

Fitness Sharing Chaotic Particle Swarm Optimization (FSCPSO): A Metaheuristic Approach for Allocating Dynamic Virtual Machine (VM) in Fog Computing Architecture

引用

computers, Materials & Continua 2024年第8期80卷 2557-2578页

作者： Prasanna Kumar Kannughatta Ranganna Siddesh Gaddadevara Matt Chin-Ling Chen Ananda Babu Jayachandra Yong-Yuan Deng Department of Computer Science&Engineering Ramaiah Institute of TechnologyBangalore560054India Department of Artificial Intelligence&Data Science Ramaiah Institute of TechnologyBangalore560054India School of Information Engineering Changchun Sci-Tech UniversityChangchun130600China Department of Computer Science and Information Engineering Chaoyang University of TechnologyTaichung41349Taiwan Department of Information Science and Engineering Malnad College of EngineeringHassan573202India

In recent decades,fog computing has played a vital role in executing parallel computational tasks,specifically,scientific workflow *** cloud data centers,fog computing takes more time to run workflow ***,it is essential to develop effective models for Virtual Machine(VM)allocation and task scheduling in fog computing *** task scheduling,VM migration,and allocation,altogether optimize the use of computational resources across different fog *** process ensures that the tasks are executed with minimal energy consumption,which reduces the chances of resource *** this manuscript,the proposed framework comprises two phases:(i)effective task scheduling using a fractional selectivity approach and(ii)VM allocation by proposing an algorithm by the name of Fitness Sharing Chaotic Particle Swarm Optimization(FSCPSO).The proposed FSCPSO algorithm integrates the concepts of chaos theory and fitness sharing that effectively balance both global exploration and local *** balance enables the use of a wide range of solutions that leads to minimal total cost and makespan,in comparison to other traditional optimization *** FSCPSO algorithm’s performance is analyzed using six evaluation measures namely,Load Balancing Level(LBL),Average Resource Utilization(ARU),total cost,makespan,energy consumption,and response *** relation to the conventional optimization algorithms,the FSCPSO algorithm achieves a higher LBL of 39.12%,ARU of 58.15%,a minimal total cost of 1175,and a makespan of 85.87 ms,particularly when evaluated for 50 tasks.

关键词： Fog computing fractional selectivity approach particle swarm optimization algorithm task scheduling virtual machine allocation

来源：评论

学校读者我要写书评

暂无评论

Flu-Net: two-stream deep heterogeneous network to detect flu like symptoms from videos using grey wolf optimization algorithm

引用

Journal of Ambient Intelligence and Humanized Computing 2023年第6期14卷 7733-7745页

作者： Gupta, Himanshu Imran, Javed Sharma, Chandani Department of Computer Science and Engineering Quantum University Roorkee India Computer Science and Engineering Department Thapar Institute of Engineering and Technology Patiala India

The outbreak of COVID-19 (also known as Coronavirus) has put the entire world at risk. The disease first appears in Wuhan, China, and later spread to other countries, taking a form of a pandemic. In this paper, we try to build an artificial intelligence (AI) powered framework called Flu-Net to identify flu-like symptoms (which is also an important symptom of Covid-19) in people, and limit the spread of infection. Our approach is based on the application of human action recognition in surveillance systems, where videos captured by closed-circuit television (CCTV) cameras are processed through state-of-the-art deep learning techniques to recognize different activities like coughing, sneezing, etc. The proposed framework has three major steps. First, to suppress irrelevant background details in an input video, a frame difference operation is performed to extract foreground motion information. Second, a two-stream heterogeneous network based on 2D and 3D Convolutional Neural Networks (ConvNets) is trained using the RGB frame differences. And third, the features extracted from both the streams are combined using Grey Wolf Optimization (GWO) based feature selection technique. The experiments conducted on BII Sneeze-Cough (BIISC) video dataset show that our framework can 70% accuracy, outperforming the baseline results by more than 8%. © 2023, The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature.

关键词： COVID-19

来源：评论

学校读者我要写书评

暂无评论

Implementation of Voice Command-based Object Detection System for Wheeled Humanoid Robot Using Deep Learning

引用

Journal of Institute of Control, Robotics and Systems 2023年第2期29卷 88-97页

作者： Jeong, Yoon-Jin Han, Ji-Hyeong Department of Computer Science and Engineering Seoul National University of Science and Technology Korea Republic of

The demand for service robots is continuously increasing. Robots that can interact with humans are required not only in factories but also in everyday life. This paper proposes a Human-Robot Interaction (HRI) system based on HUMIC, a wheeled humanoid robot with a Kinect V2 and four mecanum wheels. In the proposed system, when a person issues a voice command to an object to find, the HUMIC detects an object and moves in front of the object. For this purpose, deep learning models, i.e., wav2vec 2.0 and YOLO v3, are used for speech recognition and object detection. Furthermore, the proposed system is configured to select one object by interacting with a human when there is more than one object to find and prevents the object from disappearing from the robot’s view while moving. The experiments are conducted in a real-world environment. Because this study focuses on HRI, the experiments are conducted based on a scenario in which HUMIC detects an object in an environment without obstacles. The scenario was attempted nine times and succeeded eight times to achieve an 88% success rate. © ICROS 2023.

关键词： Human robot interaction

来源：评论

学校读者我要写书评

暂无评论

Magnetic Field-Based Reward Shaping for Goal-Conditioned Reinforcement Learning

引用

IEEE/CAA Journal of Automatica Sinica 2023年第12期10卷 2233-2247页

作者： Hongyu Ding Yuanze Tang Qing Wu Bo Wang Chunlin Chen Zhi Wang the Department of Control Science and Intelligence Engineering School of Management and EngineeringNanjing UniversityNanjing 210093China the Department of Power Engineering and Process Machinery School of Mechanical and Power EngineeringEast China University of Science and TechnologyShanghai 200237China IEEE

Goal-conditioned reinforcement learning(RL)is an interesting extension of the traditional RL framework,where the dynamic environment and reward sparsity can cause conventional learning algorithms to *** shaping is a practical approach to improving sample efficiency by embedding human domain knowledge into the learning *** reward shaping methods for goal-conditioned RL are typically built on distance metrics with a linear and isotropic distribution,which may fail to provide sufficient information about the ever-changing environment with high *** paper proposes a novel magnetic field-based reward shaping(MFRS)method for goal-conditioned RL tasks with dynamic target and *** by the physical properties of magnets,we consider the target and obstacles as permanent magnets and establish the reward function according to the intensity values of the magnetic field generated by these *** nonlinear and anisotropic distribution of the magnetic field intensity can provide more accessible and conducive information about the optimization landscape,thus introducing a more sophisticated magnetic reward compared to the distance-based ***,we transform our magnetic reward to the form of potential-based reward shaping by learning a secondary potential function concurrently to ensure the optimal policy invariance of our *** results in both simulated and real-world robotic manipulation tasks demonstrate that MFRS outperforms relevant existing methods and effectively improves the sample efficiency of RL algorithms in goal-conditioned tasks with various dynamics of the target and obstacles.

关键词： Dynamic environments goal-conditioned reinforcement learning magnetic field reward shaping

来源：评论

学校读者我要写书评

暂无评论

LMCF-FS: A Novel Lightweight Malware Classification Framework Driven by Feature Selection 6th

LMCF-FS: A Novel Lightweight Malware Classification Framewor...

引用

6th International Conference on Machine Learning for Cyber Security, ML4CS 2024

作者： Yun, Cui Zhou, Lei Xing, Shuangshuang Yang, Ning Zhao, Pan Chen, Zhiguo School of Computer Jiangsu University of Science and Technology Zhenjiang212100 China Engineering Research Center of Digital Forensics Ministry of Education Nanjing University of Information Science and Technology Nanjing210044 China School of Computer Science and School of Cyber Science and Engineering Nanjing University of Information Science and Technology Nanjing210044 China

ISBN: (纸本)9789819645657

The rapid increase in the number of malware and its variants poses a significant threat to internet security. While existing machine learning-based malware classification methods can improve accuracy, they often require complex feature engineering. In contrast, converting executable files into images and classifying them using deep learning models can reduce the reliance on prior knowledge of malware features and simplify the feature processing. However, most studies have not adequately considered the impact of image pixel features and size on classification results. In addition, the model structure of deep learning classification methods based on pre-trained parameters and transfer learning is usually more complex, resulting in the classification model requiring more memory and time overhead. To solve the above problems, this paper proposes a novel lightweight malware classification framework LMCF-FS driven by feature selection. This framework uses a proposed new feature selection method based on pixel pair features to improve feature sparsity and optimize image texture features, thereby improving feature expression capabilities. At the same time, an interpolation algorithm is used to balance the image size. In addition, a new lightweight malware classification model ConvInceptionNet is built based on the Inception-C module to balance accuracy, computational cost, and number of parameters, thereby improving malware classification efficiency. Our proposed LMCF-FS framework achieves a classification accuracy of 99.12% on the Microsoft Malware Classification Challenge Dataset (BIG2015 Dataset), and it only takes 0.92 ms to predict a 224 × 224 malware image, verifying LMCF-FS effectiveness in dealing with malware classification. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025.

关键词： Image texture

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：