检索结果-内蒙古大学图书馆

Supervised reinforcement learning based trajectory tracking control for autonomous vehicles ⁎

IFAC-PapersOnLine 2024年第10期58卷 140-145页

作者： András Mihály Van Tan Vu Trong Tu Do Péter Gáspár Systems and Control Laboratory Institute for Computer Science and Control (SZTAKI) Hungarian Research Network Kende u. 13-17 H-1111 Budapest Hungary Department of Automotive Mechanical Engineering Faculty of Mechanical Engineering University of Transport and Communications Hanoi Vietnam Mechanical and Power Engineering Faculty Electric Power University Hanoi Vietnam

Recent developments in autonomous vehicle technologies and applications gain a lot of interest by the public, as the popularity of both driver assistance and automated driving systems increase. One of the most promising aspect of the autonomous vehicle compared to conventional human driven vehicle is the increased level of safety. Machine learning techniques enables to achieve fast and efficient control actions compared to model based techniques. However, the advantages of a more conservative model based controller are their better robustness properties. In this paper a synergy of the two control philosophy is presented through a trajectory tracking control design for autonomous vehicles. A supervised reinforcement learning (RL) control method is introduced, where a robust Linear Parameter Varying (LPV) controller supervises the operation of the trained RL agent. Thus, in case sensor noise is detected, the guaranteed stability LPV controller takes over the steering control action. In order to demonstrate the operation of the proposed method, three different simulations have been evaluated and compared in CarSim simulation environment.

关键词： autonomous vehicle control supervised reinforcement learning trajectory tracking

来源：评论

学校读者我要写书评

暂无评论

Attention-Based Difficulty Feature Enhancement for Knowledge Tracing

Attention-Based Difficulty Feature Enhancement for Knowledge...

引用

International Joint Conference on Neural networks (IJCNN)

作者： Xinhua Wang Hao Lu Liancheng Xu Lei Guo Xiaohui Zhao School of Information Science and Engineering Shandong Normal University Jinan China School of Mathematics and Statistics Shandong Normal University Jinan China Shandong Key Laboratory of Computer Network Jinan China

ISBN: (数字)9798350359312

ISBN: (纸本)9798350359329

The task of knowledge tracing aims to monitor students’ knowledge states through their historical answer records and predict their future performance in answering questions. In recent years, knowledge tracing models based on deep learning have exhibited significantly higher accuracy in prediction compared to traditional knowledge tracing models. However, existing models often fail to fully leverage the impact of difficulty factors on students’ knowledge states. In this paper, to better exploit the difficulty factors for enhancing the discriminative power among different questions, we propose a novel Difficulty-Fusion Knowledge Tracing(DFAKT) model. Specifically, we design a DEI(Difficult-Enhance Interaction) method for extracting difficulty features by integrating skill difficulty and question-specific difficulty information. Subsequently, we develop a Difficulty-Fusion module designed to integrate the student interaction information, which includes features such as answers and knowledge points, with the difficulty feature. Finally, recognizing that different exercises have varying impacts on students’ knowledge states, we incorporate an attention-based GRU module in the model to dynamically aggregate the knowledge states from previous time steps. This allows the model to focus more on questions with a greater impact on knowledge states, thereby improving prediction accuracy. Experimental results demonstrate that the proposed model outperforms existing models in predictive performance on large-scale real-world datasets, and the effectiveness of each module is validated.

关键词： Knowledge engineering Deep learning Accuracy Neural networks Focusing Predictive models Feature extraction

来源：评论

学校读者我要写书评

暂无评论

ConvCatb: An Attention-based CNN-CATBOOST Risk Prediction Model for Driving Safety

ConvCatb: An Attention-based CNN-CATBOOST Risk Prediction Mo...

引用

2021 International Conference on networking and network Applications, NaNA 2021

作者： Hei, Xinhong Zhang, Hao Ji, Wenjiang Wang, Yichuan Zhu, Lei Qiu, Yuan College of Computer Science and Engineering Xi'an University of Technology Xi'an China Shaanxi Key Laboratory for Network Computing and Security Technology Xi'an China

ISBN: (纸本)9781665441582

Risk prediction is one of the most important tasks in assistant and automatic driving. In recent years, by the help of VANET and various sensors in the cars, the status of cars and roads can be collected in real time and used for data-driven based driving risk prediction. However, it is challenging to predict the driving risk due to the complex relationship between multiple environment factors like location, weather, time etc. Thus, a deep learning model ConvCatb was proposed in this paper, which improves the attention mechanism to the traditional Convolutional Neural networks, and combines the CatBoost algorithm to predict the current driving safety. The main idea is to emphasize the combination relationship between driving environment features through Non-local attention mechanism, and then use CatBoost to replace the softmax of CNN for classification. Finally, the experiment results show that the ConvCatb achieved superiorities in accuracy and F1-score, compared with existing schemes. © 2021 IEEE.

关键词： Automobile drivers

来源：评论

学校读者我要写书评

暂无评论

A fine-grained petri model for SQL time-blind injection

A fine-grained petri model for SQL time-blind injection

引用

2021 International Conference on networking and network Applications, NaNA 2021

作者： Zhang, Guiling Zhang, Yaling Wang, Yichuan Zhu, Lei Ji, Wenjiang Xi'an University of technology School of computer science and engineering Xi'an China Shaanxi Key Laboratory for Network Computing and Security Technology Xi'an China

ISBN: (纸本)9781665441582

With the background of severe SQL injection attacks, the existing SQL injection modeling methods cannot reflect the process of SQL injection attacks in a fine-grained manner. Based on the discussion of attack technology, this paper takes SQL time-blind injection as an example to model its process with Petri Net. The validity of the model is verified by quantitative analysis and qualitative analysis. Try to inject 10, 20, 30, 40 and 50 times into target aircraft and Petri Net model respectively. The blind injection time is recorded and compared. The results show that the injection time increases with the increase of injection times. Under the same injection times, the Petri Net model takes less time. The sending time in the token can be set. When the sending time is short, the injection speed is fast, and super real-time simulation can be realized, which can realize the rapid prediction of attacks and resource vulnerability effects. When the sending time is long, the injection process slows down. It is beneficial to observe the details of the injection process and whether conflicts occur at a fine-grained level, analyze the purpose of the attack and achieve the purpose of building a patch model. The patch model can effectively take countermeasures against attacks, predict unknown vulnerabilities and ensure network information security. © 2021 IEEE.

关键词： Petri nets

来源：评论

学校读者我要写书评

暂无评论

Audio-Visual Generalized Zero-Shot Learning Based on Variational Information Bottleneck

Audio-Visual Generalized Zero-Shot Learning Based on Variati...

引用

IEEE International Conference on Multimedia and Expo (ICME)

作者： Yapeng Li Yong Luo Bo Du School of Computer Science National Engineering Research Center for Multimedia Software Institute of Artificial Intelligence and Hubei Key Laboratory of Multimedia and Network Communication Engineering Wuhan University Wuhan China

Audio-visual generalized zero-shot learning (GZSL) aims to train a model on seen classes for classifying data samples from both seen classes and unseen classes. Due to the absence of unseen training samples, the model tends to misclassify unseen class samples into seen classes. To mitigate this problem, in this paper, we propose a method based on variational information bottleneck for audio-visual GZSL. Specifically, we model the joint representations as a product-of-experts over marginal representations to integrate the information of audio and visual. Besides, we introduce variational information bottleneck to the learning of audio-visual joint representations and marginal representations of audio, visual, and text label modalities. This helps our model reduce the negative impact of information that cannot be generalized to unseen classes. Experimental results conducted on the UCF-GZSL, VGGSound-GZSL, and ActivityNet-GZSL benchmarks demonstrate the effectiveness and superiority of the proposed model for audio-visual GZSL.

关键词：

来源：评论

学校读者我要写书评

暂无评论

The Entrapment Problem in Random Walk Decentralized Learning

arXiv

引用

arXiv 2024年

作者： Liu, Zonghong Rouayheb, Salim El Dwyer, Matthew Department of Electrical and Computer Engineering Rutgers University New BrunswickNJ United States Network Cyber and Computational Sciences Division DEVCOM Army Research Laboratory AdelphiMD United States

This paper explores decentralized learning in a graph-based setting, where data is distributed across nodes. We investigate a decentralized SGD algorithm that utilizes a random walk to update a global model based on local data. Our focus is on designing the transition probability matrix to speed up convergence. While importance sampling can enhance centralized learning, its decentralized counterpart, using the Metropolis-Hastings (MH) algorithm, can lead to the entrapment problem, where the random walk becomes stuck at certain nodes, slowing convergence. To address this, we propose the Metropolis-Hastings with Lévy Jumps (MHLJ) algorithm, which incorporates random perturbations (jumps) to overcome entrapment. We theoretically establish the convergence rate and error gap of MHLJ and validate our findings through numerical experiments. Copyright © 2024, The Authors. All rights reserved.

关键词： Contrastive Learning

来源：评论

学校读者我要写书评

暂无评论

Monocular Camera-Based Visual Truck Type Recognition

SSRN

引用

SSRN 2024年

作者： Luo, Haibo Ye, Huaisheng Feng, Chao Shi, Xin Xie, Guoqi School of Computer and Data Science Minjiang University Fuzhou350018 China College of Computer and Data Science Fuzhou University Fuzhou350018 China College of Network and Information Security Fujian Agriculture and Forestry University Fuzhou350018 China Key Laboratory for Embedded and Network Computing of Hunan Province College of Computer Science and Electronic Engineering Hunan University Changsha410082 China College of Computer Science and Electronic Engineering Hunan University Changsha410082 China

Truck overload and over-limit are the primary causes of infrastructure damage and traffic safety accidents. In the past 2 years, researchers have started to deploy intelligent Internet of Things system at the source of truck delivery to enforce permit control. Truck type recognition is the core of such a system. However, existing methods suffer from limitations such as low recognition accuracy, high costs, difficult maintenance, and the need for road surface modifications. This study proposes a monocular camera-based visual truck type recognition method. First, the optimized lightweight deep neural network based on machine vision is utilized to detect truckaxles and recognize wheel hub categories. Next, a tracking algorithm and the video streamcontext information are employed to complete the robust statistics of the axle by neighborhoodcounting. Finally, the proposed method further achieves truck type recognition by leveragingavailable prior knowledge. Experimental results demonstrate that this method exhibits superiortime efficiency compared with existing algorithms, and it reduces average processing time by92.37% to 96.24% while ensuring high accuracy. Moreover, the method has been deployed onembedded devices with low computing power to enable real-time operation onsite. © 2024, The Authors. All rights reserved.

关键词： Object detection

来源：评论

学校读者我要写书评

暂无评论

Towards Modality-Agnostic Person Re-identification with Descriptive Query

Towards Modality-Agnostic Person Re-identification with Desc...

引用

Conference on computer Vision and Pattern Recognition (CVPR)

作者： Cuiqun Chen Mang Ye Ding Jiang National Engineering Research Center for Multimedia Software Hubei Key Laboratory of Multimedia and Network Communication Engineering Institute of Artificial Intelligence School of Computer Science Wuhan University Wuhan China Hubei Luojia Laboratory Wuhan China

Person re-identification (ReID) with descriptive query (text or sketch) provides an important supplement for general image-image paradigms, which is usually studied in a single cross-modality matching manner, e.g., text-to-image or sketch-to-photo. However, without a camera-captured photo query, it is uncertain whether the text or sketch is available or not in practical scenarios. This motivates us to study a new and challenging modality-agnostic person re-ideruification problem. Towards this goal, we propose a unified person re-identification (UNIReID) architecture that can effectively adapt to cross-modality and multi-modality tasks. Specifically, UNIReID incorporates a simple dual-encoder with task-specific modality learning to mine and fuse visual and textual modality information. To deal with the imbalanced training problem of different tasks in UNIReID, we propose a task-aware dynamic training strategy in terms of task difficulty, adaptively adjusting the training focus. Besides, we construct three multi-modal ReID datasets by collecting the corresponding sketches from photos to support this challenging study. The experimental results on three multi-modal ReID datasets show that our UNIReID greatly improves the retrieval accuracy and generalization ability on different tasks and unseen scenarios.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Unsupervised Visible-Infrared Person Re-Identification via Progressive Graph Matching and Alternate Learning

Unsupervised Visible-Infrared Person Re-Identification via P...

引用

Conference on computer Vision and Pattern Recognition (CVPR)

作者： Zesen Wu Mang Ye Hubei Key Laboratory of Multimedia and Network Communication Engineering National Engineering Research Center for Multimedia Software Institute of Artificial Intelligence School of Computer Science Wuhan University Wuhan China Hubei Luojia Laboratory Wuhan China

Unsupervised visible-infrared person re-identification is a challenging task due to the large modality gap and the unavailability of cross-modality correspondences. Cross-modality correspondences are very crucial to bridge the modality gap. Some existing works try to mine cross-modality correspondences, but they focus only on local information. They do not fully exploit the global relationship across identities, thus limiting the quality of the mined correspondences. Worse still, the number of clusters of the two modalities is often inconsistent, exacerbating the unreliability of the generated correspondences. In response, we devise a Progressive Graph Matching method to globally mine cross-modality correspondences under cluster imbalance scenarios. PGM formulates correspondence mining as a graph matching process and considers the global information by minimizing the global matching cost, where the matching cost measures the dissimilarity of clusters. Besides, PGM adopts a progressive strategy to address the imbalance issue with multiple dynamic matching processes. Based on PGM, we design an Alternate Cross Contrastive Learning (ACCL) module to reduce the modality gap with the mined cross-modality correspondences, while mitigating the effect of noise in correspondences through an alternate scheme. Extensive experiments demonstrate the reliability of the generated correspondences and the effectiveness of our method.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Freevc: Towards High-Quality Text-Free One-Shot Voice Conversion

Freevc: Towards High-Quality Text-Free One-Shot Voice Conver...

引用

International Conference on Acoustics, Speech, and Signal Processing (ICASSP)

作者： Jingyi Li Weiping Tu Li Xiao National Engineering Research Center for Multimedia Software School of Computer Science Wuhan University Wuhan China Hubei Key Laboratory of Multimedia and Network Communication Engineering Wuhan University Wuhan China Hubei Luojia Laboratory China

Voice conversion (VC) can be achieved by first extracting source content information and target speaker information, and then reconstructing waveform with these information. However, current approaches normally either extract dirty content information with speaker information leaked in, or demand a large amount of annotated data for training. Besides, the quality of reconstructed waveform can be degraded by the mismatch between conversion model and vocoder. In this paper, we adopt the end-to-end framework of VITS for high-quality waveform reconstruction, and propose strategies for clean content information extraction without text annotation. We disentangle content information by imposing an information bottleneck to WavLM features, and propose the spectrogram-resize based data augmentation to improve the purity of extracted content information. Experimental results show that the proposed method outperforms the latest VC models trained with annotated data and has greater robustness.

关键词： Training Annotations Vocoders Signal processing Feature extraction Information retrieval Data models

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：