检索结果-内蒙古大学图书馆

IEEE International Conference on Big Data

作者： Fredy Rojas Samaneh Madanian John Michael Templeton Christian Poellabauer Sandra L. Schneider Department of Computer Science and Software Engineering Auckland University of Technology (AUT) Auckland New Zealand Department of Computer Science and Engineering University of South Florida Tampa FL USA School of Computing and Information Sciences Florida International University Miami FL USA Department of Communicative Sciences & Disorders Saint Mary’s College Notre Dame IN USA

ISBN: (数字)9798350362480

ISBN: (纸本)9798350362497

Mild traumatic brain injury (mTBI) is challenging to diagnose due to its subtle and transient symptoms, making noninvasive diagnostic tools crucial for early detection. This study explores the use of a custom ResNet deep learning model combined with the Grad-CAM interpretability technique for mTBI detection via speech analysis. Speech data were transformed into Mel-spectrograms and fed into the model for binary classification between concussed and control individuals. The Grad-CAM method provided insights into which frequency regions of the Mel-spectrogram were most important for the model's predictions, with higher-frequency regions identified as significant for the model in detecting mTBI. Using Monte Carlo Cross-Validation (MCCV), we evaluated 50 different subject train-test split configurations to gain insights into the model's performance stability and variability. This analysis can assess the model's ability to learn consistent patterns within the dataset and suggest potential generalization tendencies. The variability observed in the performance metrics distribution underscores the importance of robust evaluation methods, particularly when working with small datasets. The combination of deep learning, robust evaluation technique and interpretability in this study contributes to the development of clinically viable, speech-based tools for mTBI detection, with potential applications in sports and healthcare settings.

关键词： Deep learning Analytical models Medical services Predictive models Brain modeling Data models Stability analysis Transient analysis Brain injuries Sports

来源：评论

学校读者我要写书评

暂无评论

Understanding the Robustness of 3D Object Detection with Bird'View Representations in Autonomous Driving

Understanding the Robustness of 3D Object Detection with Bir...

引用

2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2023

作者： Zhu, Zijian Zhang, Yichi Chen, Hai Dong, Yinpeng Zhao, Shu Ding, Wenbo Zhong, Jiachen Zheng, Shibao Institute of Image Communication and Network Engineering Shanghai Jiao Tong University China Institute for Ai Tsinghua University BNRist Center Thbi Lab Dept. of Comp. Sci. and Tech. China School of Computer Science and Technology Anhui University Key Laboratory of Intelligent Computing and Signal Processing Ministry of Education Information Materials and Intelligent Sensing Laboratory of Anhui Province China Saic Motor Ai Lab Zhongguancun Laboratory China

ISBN: (纸本)9798350301298

3D object detection is an essential perception task in autonomous driving to understand the environments. The Bird's-Eye-View (BEV) representations have significantly improved the performance of 3D detectors with camera inputs on popular benchmarks. However, there still lacks a systematic understanding of the robustness of these vision-dependent BEV models, which is closely related to the safety of autonomous driving systems. In this paper, we evaluate the natural and adversarial robustness of various representative models under extensive settings, to fully understand their behaviors influenced by explicit BEV features compared with those without BEV. In addition to the classic settings, we propose a 3D consistent patch attack by applying adversarial patches in the 3D space to guarantee the spatiotemporal consistency, which is more realistic for the scenario of autonomous driving. With substantial experiments, we draw several findings: 1) BEV models tend to be more stable than previous methods under different natural conditions and common corruptions due to the expressive spatial representations;2) BEV models are more vulnerable to adversarial noises, mainly caused by the redundant BEV features;3) Camera-LiDARfusion models have superior performance under different settings with multi-modal inputs, but BEV fusion model is still vulnerable to adversarial noises of both point cloud and image. These findings alert the safety issue in the applications of BEV detectors and could facilitate the development of more robust models. © 2023 IEEE.

关键词： Autonomous driving

来源：评论

学校读者我要写书评

暂无评论

Artificial Neural Network (ANN) With Chan-Vese(CV) Algorithm-Based Plant Disease Detection And Classification

Artificial Neural Network (ANN) With Chan-Vese(CV) Algorithm...

引用

Applied Intelligence and Sustainable computing (ICAISC), International Conference on

作者： Avr Mayuri V Rashmi Bappaditya Das Amitabha Mandal S. Neelakandan School of Computing Science and Engineering VIT Bhopal University Bhopal India Department of Information Technology Prasad V. Potluri Siddhartha Institute of Technology Vijayawada India Department of computer science & engineering Dr. B. C. Roy engineering college Durgapur India Department of computer science & engineering R.M.K Engineering College Kavaraipettai India

A country’s economy is influenced by a wide range of factors, with the agriculture, service, and food processing sectors. In our country, agriculture is one of the most significant economic sectors. The agriculture sector, however, has several difficulties, such as the different climatic conditions in other regions of our country that give rise to various infectious diseases in different plant species, resulting in a significant drop in crop productivity and income generation. Early and accurate plant disease diagnosis is crucial in improving the foundation for harvest formation. The traditional method of plant disease detection through eye observation is unsuccessful since it necessitates knowledge. Machine learning (ML) techniques are being utilized to identify and categorize plant ailments to address this issue. In this work, plant diseases can be classified using machine learning methods. These involve classifier training, classification, and stages like dataset construction, image loading, pre-processing, segmentation, and feature extraction. An artificial neural network is initially used to distinguish and find leaves in a challenging environment. The Chan-Vese (CV) technique states that symptoms can be found in images that have been segmented depending on the outcomes of the ANN algorithm. A transfer learning model trained on a dataset of ill leaves on an essential background is then given the segmented leaves. The model is also checked for rust, black rot, and bacterial plaque conditions. The findings demonstrate that the methodology is more accurate—98.22%—than the previous approach, minimizing the adverse effects of illness on agricultural productivity and promoting long-term agrarian growth. Utilizing data from PlantVillage, the model is trained and validated. The dataset includes images of plant diseases afflicting pepper, tomato, and potato plants. Fifteen different plant disease categories are used to train and evaluate the classifier. The results of t

关键词：

来源：评论

学校读者我要写书评

暂无评论

A Comprehensive Review of Multimodal Large Language Models: Performance and Challenges Across Different Tasks

arXiv

引用

arXiv 2024年

作者： Wang, Jiaqi Jiang, Hanqi Liu, Yiheng Ma, Chong Zhang, Xu Pan, Yi Liu, Mengyuan Gu, Peiran Xia, Sichen Li, Wenjun Zhang, Yutong Wu, Zihao Liu, Zhengliang Zhong, Tianyang Ge, Bao Zhang, Tuo Qiang, Ning Hu, Xintao Jiang, Xi Zhang, Xin Zhang, Wei Shen, Dinggang Liu, Tianming Zhang, Shu The School of Computer Science Northwestern Polytechnical University Xi’an710072 China The School of Automation Northwestern Polytechnical University Xi’an710072 China The Institute of Medical Research Northwestern Polytechnical University Xi’an710072 China The School of Computing The University of Georgia Athens30602 United States The School of Physics and Information Technology Shaanxi Normal University Xi’an710119 China The Clinical Hospital of Chengdu Brain Science Institute MOE Key Lab for Neuroinformation School of Life Science and Technology University of Electronic Science and Technology of China Chengdu China The School of Computer and Cyber Sciences Augusta University Augusta30912 United States The School of Biomedical Engineering Shang-haiTech University Shanghai201210 China Shanghai United Imaging Intelligence Co. Ltd. Shanghai200230 China Shanghai Clinical Research and Trial Center Shanghai201210 China

In an era defined by the explosive growth of data and rapid technological advancements, Multimodal Large Language Models (MLLMs) stand at the forefront of artificial intelligence (AI) systems. Designed to seamlessly integrate diverse data types—including text, images, videos, audio, and physiological sequences—MLLMs address the complexities of real-world applications far beyond the capabilities of single-modality systems. In this paper, we systematically sort out the applications of MLLM in multimodal tasks such as natural language, vision, and audio. We also provide a comparative analysis of the focus of different MLLMs in the tasks, and provide insights into the shortcomings of current MLLMs, and suggest potential directions for future research. Through these discussions, this paper hopes to provide valuable insights for the further development and application of MLLM. Copyright © 2024, The Authors. All rights reserved.

关键词： Natural language processing systems

来源：评论

学校读者我要写书评

暂无评论

HairDiffusion: Vivid Multi-Colored Hair Editing via Latent Diffusion

arXiv

引用

arXiv 2024年

作者： Zeng, Yu Zhang, Yang Liu, Jiachen Shen, Linlin Deng, Kaijun He, Weizhao Wang, Jinbao Computer Vision Institute School of Computer Science & Software Engineering Shenzhen University China Shenzhen Institute of Artificial Intelligence and Robotics for Society China National Engineering Laboratory for Big Data System Computing Technology Shenzhen University China Guangdong Provincial Key Laboratory of Intelligent Information Processing China

Hair editing is a critical image synthesis task that aims to edit hair color and hairstyle using text descriptions or reference images, while preserving irrelevant attributes (e.g., identity, background, cloth). Many existing methods are based on StyleGAN to address this task. However, due to the limited spatial distribution of StyleGAN, it struggles with multiple hair color editing and facial preservation. Considering the advancements in diffusion models, we utilize Latent Diffusion Models (LDMs) for hairstyle editing. Our approach introduces Multi-stage Hairstyle Blend (MHB), effectively separating control of hair color and hairstyle in diffusion latent space. Additionally, we train a warping module to align the hair color with the target region. To further enhance multi-color hairstyle editing, we fine-tuned a CLIP model using a multi-color hairstyle dataset. Our method not only tackles the complexity of multi-color hairstyles but also addresses the challenge of preserving original colors during diffusion editing. Extensive experiments showcase the superiority of our method in editing multi-color hairstyles while preserving facial attributes given textual descriptions and reference images. © 2024, CC BY.

关键词： Color image processing

来源：评论

学校读者我要写书评

暂无评论

Shortest Path Priority-based RPL (SPPB-RPL): The Case of a Smart Campus

Shortest Path Priority-based RPL (SPPB-RPL): The Case of a S...

引用

2020 IEEE Conference on Application, information and Network Security, AINS 2020

作者： Onwuegbuzie, Innocent Uzougbo Razak, Shukor Abd Isnin, Ismail Fauzi Anuar, Nor Badrul School of Computing Faculty of Engineering Johor Malaysia Faculty of Computer Science and Information Technology University of Malaya Department of Computer System and Technology Kuala Lumpur Malaysia

ISBN: (纸本)9781728192406

Wireless Sensor Networks (WSN) is the network of embedded and resource-constrained nodes, that forms the foundation of the Internet of Things (IoT). These networks are designed to sense, aggregate and transport sensed data to locations that may be local or remote to the sensory field, for immediate or stored for later use. In most cases, WSN generates heterogeneous data that conveys information of varying priorities which may be real-time (RT) or non-real-time (NRT). However, the de facto routing protocol;Routing Protocol for Low-power and Lossy Networks (RPL) is not optimized for heterogeneous data routing. Likewise, routing schemes proposed in the literature have not proffered a lasting solution. In this paper, we proposed the Shortest Path Priority-based RPL (SPPB-RPL). The SPPB-RPL scheme not only routes data with varying priorities but also does so with optimal network performances. Its application in a smart campus is used as a case study. The scheme uses the mechanism of class of service (CS) to classify and prioritized data into various priority classes. Subsequently, the standard RPL DIO base object was improved upon to accommodate the prioritized data classes, and finally, the Dijkstra Algorithm (DA) was adopted to generate the optimal shortest routing path leading to the destination node which is guided by the priority of the data. Benchmarked with related schemes, SPPB-RPL performed optimally with a 99.98% packet delivery ratio with the least latency. This enhanced the network performance for a prolonged operational lifetime. © 2020 IEEE.

关键词： Routing protocols

来源：评论

学校读者我要写书评

暂无评论

CA-Edit: Causality-Aware Condition Adapter for High-Fidelity Local Facial Attribute Editing

arXiv

引用

arXiv 2024年

作者： Xian, Xiaole He, Xilin Niu, Zenghao Zhang, Junliang Xie, Weicheng Song, Siyang Yu, Zitong Shen, Linlin Computer Vision Institute School of Computer Science & Software Engineering Shenzhen University China National Engineering Laboratory for Big Data System Computing Technology Shenzhen University China Guangdong Key Laboratory of Intelligent Information Processing China University of Exeter United Kingdom Great Bay University China

For efficient and high-fidelity local facial attribute editing, most existing editing methods either require additional finetuning for different editing effects or tend to affect beyond the editing regions. Alternatively, inpainting methods can edit the target image region while preserving external areas. However, current inpainting methods still suffer from the generation misalignment with facial attributes description and the loss of facial skin details. To address these challenges, (i) a novel data utilization strategy is introduced to construct datasets consisting of attribute-text-image triples from a data-driven perspective, (ii) a Causality-Aware Condition Adapter is proposed to enhance the contextual causality modeling of specific details, which encodes the skin details from the original image while preventing conflicts between these cues and textual conditions. In addition, a Skin Transition Frequency Guidance technique is introduced for the local modeling of contextual causality via sampling guidance driven by low-frequency alignment. Extensive quantitative and qualitative experiments demonstrate the effectiveness of our method in boosting both fidelity and editability for localized attribute editing. The code is available at https://***/connorxian/CA-Edit. © 2024, CC BY.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Foreign Material Detection and Localization in Stored Grain with Passive RFID Tag Arrays

Foreign Material Detection and Localization in Stored Grain ...

引用

IEEE International Conference on RFID-technology and Applications (RFID-TA)

作者： Erbo Shen Weidong Yang Xuyu Wang Shiwen Mao Henan Key Laboratory of Grain Photoelectric Detection & Control Henan University of Technology Zhengzhou China Kaifeng University Kaifeng China College of Information Science & Engineering Henan University of Technology Zhengzhou China Knight Foundation School of Computing & Information Sciences Florida International University Miami FL USA Department of Electrical and Computer Engineering Auburn University Auburn AL USA

In the early stage of harvesting, transportation, and storage, the grain could be mingled with clods from the field, metal pieces from aging machines, and other objects of foreign material, which will greatly influence the grain quality and food security. In this paper, we propose a novel radio frequency (RF) sensing system termed TagSee, which leverages passive RFID tag arrays. The goal is to simultaneously sense the presence of foreign materials and locate their locations in the 3D space, to automatically monitor the stored grain. Specifically, we use RFID received signal strength (RSS) and phase as features for foreign material detection. To design the TagSee system, we first introduce a sensing space division method. Then, an Euclidean Distance Ratio (EDR) algorithm and a heuristic method are proposed to achieve high localization accuracy. Experimental results show that TagSee can effectively detect foreign materials in stored grain and achieve a centimeter-level localization accuracy.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Meta-Learning Strategies for Engineering Education Reform: Leveraging Convolutional Neural Networks and Reinforcement Learning Perspectives

Meta-Learning Strategies for Engineering Education Reform: L...

引用

Green Industrial Electronics and Sustainable Technologies (GIEST), IEEE International Conference on

作者： SK. Fakruddin Babavali K. Kavitha S. Suma Christal Mary Shamim Ahmad Khan Prasanta Parida B Kiran Bala Department of Physics Velagapudi Ramakrishna Siddhartha Engineering College Deemed To Be University Vijayawada Andhra Pradesh India Department of AI and ML School of Computing Mohan Babu University Tirupati India Department of Information Technology Panimalar Engineering College Chennai India Glocal School of Science & Technology Glocal University U.P India School of Rural Management KIIT University Bhubaneswar Odisha India Department of AI & DS K. Ramakrishnan College of Engineering Trichy India

ISBN: (数字)9798331540142

ISBN: (纸本)9798331540159

There are many barriers in the use of technological education like low inclusive learning education resources; better ways of educating students and outcome assessment technology -based teaching and learning approach also poses a challenge in education since traditional teaching designs meet the demands in a way that enhances poor students’ attendance in class as well as poor performance. Inasmuch as the conventional approaches for curriculum development as well as manual grading approaches are rigid and cannot be adapted to other settings. This paper introduces a new meta-learning strategy to enhance a variety of machine learning; convolutional neural networks to network visual learning aids, grading tools, and instructional information to set up learning environments. RL algorithms are used in creating student learning paths or programmers, and smart tutoring systems that adapt with the students’ performance and behavioral characteristics. Briefly, the Meta-learning strategies such as MAML and few-shot learning improve the flexibility of these models and make them capable of Rapid Adaptability to perform new learning tasks with a small amount of data. The approach consists of recognizing educational problems, obtaining and preprocessing data, choosing the suitable CNN and RL models, using meta-learning approaches. Ongoing monitoring and feedback are included with the tools in order to maintain update and modify the tools as needed. Evaluation of the performance shows a substantial enhancement in engagement of students, adoption of learning, and grading with the accuracy of 92% for AAC, precision of 90%, recall of 88%, and f1-score are 89%.

关键词： Metalearning Adaptation models Visualization Neural networks Reinforcement learning Manuals Data models Convolutional neural networks Engineering education Monitoring

来源：评论

学校读者我要写书评

暂无评论

Activity Based Easy Learning Of PushDown Automata

Activity Based Easy Learning Of PushDown Automata

引用

Cyber Resilience (ICCR), International Conference on

作者： Liaqat Ali Nizar Sahawneh Yousuf Agha Saira Kanwal Imran Nouh Sabri Elmitwally College of Engineering and Technology University of Science and Technology of Fujairah Fujairah UAE School of Business Skyline University College Sharjah UAE Faculty of information technology Monash University Department of Computer Science Agriculture University Faisalabad Pakistan Fatima memorial hospital college of medicine and dentistry Lahore Lahore Pakistan School of Computing and Digital Technology Birmingham City University Birmingham UK

ISBN: (数字)9781665461221

ISBN: (纸本)9781665461238

Teaching is a skill. It is significant to be aware of or enquiring how to express this skill. Teacher must have fully grasped his art of teaching. He must know the psychology, mental and emotional distorting elements of students to deal with in all situations accordingly. Learning disability is not a big difficulty or harder-ship and can achieve easily by developing interest and motivating students by introducing some new updated tools and methodologies like Thoth, SELFA, FLUTE, JFLAP, FLAP, Java Finite Automata, Deus Ex Machina (DEM) and homework exercises for practicing by hand, workshops, oral assessments, quizzes, group sharing, group assessment, clustering and feedback respectively that are explain below. This article provides different ways of teaching PDA and makes students' learning process easy. Furthermore this article also clarifies the conceptual model of PDA and enhances the ability to design PDA machines conveniently.

关键词： Java Art Learning automata Conferences Computational modeling Education Psychology

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：