检索结果-内蒙古大学图书馆

Asia-Pacific Signal and Information Processing Association Annual Summit and Conference (APSIPA)

作者： Bang Zeng Hongbin Suo Yulong Wan Ming Li Data Science Research Center Duke Kunshan University Kunshan China School of Computer Science Wuhan University Wuhan China Data&AI Engineering System OPPO Beijing China

Most multi-channel speaker extraction schemes use the target speaker’s location information as a reference, which must be known in advance or derived from visual cues. In addition, memory and computation costs are enormous when the model deals with the fusion input. In this paper, we propose Speaker-extraction-and-filter Network (SeafNet), which is a low-complexity multi-channel speaker extraction network with only speech cues. Specifically, the SeafNet separates the mixture by utilizing the correlation between an estimation of target speaker on reference channel and the mixed input on rest channels. Experimental results show that compared with the baseline, the SeafNet model achieves 6.4% relative SISNRi improvement on the fixed geometry array and 8.9% average relative SISNRi improvement on the ad-hoc array. Meanwhile, the SeafNet achieves 60% relative reduction in the number of parameters and 42% relative reduction in the computational cost.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Internet of Things- Cloud Security Automation Technology Based on Artificial Intelligence 1

Internet of Things- Cloud Security Automation Technology Bas...

引用

1st International Conference on Applied Artificial Intelligence and Computing, ICAAIC 2022

作者： Hussain, Karrar Vanathi, D. Jose, Bibin K Kavitha, S. Rane, Bhuvaneshwari Yogesh Kaur, Harpreet Sandhya, C. Cvr College of Engineering Ibrahimpatnam Department of Ece Telangana India Nandha Engineering College Department of Artificial Intelligence and Data Science Erode India Department of Mathematics Kerala India Saveetha College of Liberal Arts and Science Saveetha Ins Titute of Medical and Technical Sciences Department of Computer Science Chennai India KYDSCT's College of Pharmacy Pharmaceutical Chemistry Department India Lovely Professional University School of Computer Science & Engineering Punjab Phagwara India S.A.Engineering College Management of Science Chennai India

ISBN: (纸本)9781665497107

The development of industrial robots, as a carrier of artificial intelligence, has played an important role in promoting the popularisation of artificial intelligence super automation technology. The paper introduces the system structure, hardware structure, and software system of the mobile robot climber based on computer big data technology, based on this research background. At the same time, the paper focuses on the climber robot's mechanism compound method and obstacle avoidance control algorithm. Smart home computing focuses on 'home' and brings together related peripheral industries to promote smart home services such as smart appliances, home entertainment, home health care, and security monitoring in order to create a safe, secure, energy-efficient, sustainable, and comfortable residential living environment. It's been twenty years. There is still no clear definition of 'intelligence at home,' according to Philips Inc., a leading consumer electronics manufacturer, which once stated that intelligence should comprise sensing, connectedness, learning, adaption, and ease of interaction. S mart applications and services are still in the early stages of development, and not all of them can yet exhibit these five intelligent traits. © 2022 IEEE.

关键词： Industrial robots

来源：评论

学校读者我要写书评

暂无评论

Imitation learning from purified demonstrations 24

Imitation learning from purified demonstrations

引用

Proceedings of the 41st International Conference on Machine Learning

作者： Yunke Wang Minjing Dong Yukun Zhao Bo Du Chang Xu School of Computer Science National Engineering Research Center for Multimedia Software Institute of Artificial Intelligence and Wuhan Institute of Data Intelligence Wuhan University China Department of Computer Science City University of Hong Kong China School of Computer Science Faculty of Engineering The University of Sydney Australia

Imitation learning has emerged as a promising approach for addressing sequential decision-making problems, with the assumption that expert demonstrations are optimal. However, in real-world scenarios, most demonstrations are often imperfect, leading to challenges in the effectiveness of imitation learning. While existing research has focused on optimizing with imperfect demonstrations, the training typically requires a certain proportion of optimal demonstrations to guarantee performance. To tackle these problems, we propose to purify the potential noises in imperfect demonstrations first, and subsequently conduct imitation learning from these purified demonstrations. Motivated by the success of diffusion model, we introduce a two-step purification via diffusion process. In the first step, we apply a forward diffusion process to smooth potential noises in imperfect demonstrations by introducing additional noise. Subsequently, a reverse generative process is utilized to recover the optimal demonstration from the diffused ones. We provide theoretical evidence supporting our approach, demonstrating that the distance between the purified and optimal demonstration can be bounded. Empirical results on MuJoCo and RoboSuite demonstrate the effectiveness of our method from different aspects.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Label Information Bottleneck for Label Enhancement

Label Information Bottleneck for Label Enhancement

引用

Conference on computer Vision and Pattern Recognition (CVPR)

作者： Qinghai Zheng Jihua Zhu Haoyu Tang College of Computer and Data Science Fuzhou University China School of Software Engineering Xi'an Jiaotong University Xi'an China School of Software Shandong University Jinan China

In this work, we focus on the challenging problem of Label Enhancement (LE), which aims to exactly recover label distributions from logical labels, and present a novel Label Information Bottleneck (LIB) method for LE. For the recovery process of label distributions, the label irrelevant information contained in the dataset may lead to unsatisfactory recovery performance. To address this limitation, we make efforts to excavate the essential label relevant information to improve the recovery performance. Our method formulates the LE problem as the following two joint processes: 1) learning the representation with the essential label relevant information, 2) recovering label distributions based on the learned representation. The label relevant information can be excavated based on the “bottleneck” formed by the learned representation. Significantly, both the label relevant information about the label assignments and the label relevant information about the label gaps can be explored in our method. Evaluation experiments conducted on several benchmark label distribution learning datasets verify the effectiveness and competitiveness of LIB. Our source codes are available at https://***/qinghai-zheng/LIBLE.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Point Clouds Are Specialized Images: A Knowledge Transfer Approach for 3D Understanding

arXiv

引用

arXiv 2023年

作者： Kang, Jiachen Jia, Wenjing He, Xiangjian Lam, Kin Man School of Electrical and Data Engineering University of Technology Sydney Australia School of Computer Science University of Nottingham Ningbo China Department of Electrical and Electronic Engineering The Hong Kong Polytechnic University Hong Kong

Self-supervised representation learning (SSRL) has gained increasing attention in point cloud understanding, in addressing the challenges posed by 3D data scarcity and high annotation costs. This paper presents PCExpert, a novel SSRL approach that reinterprets point clouds as "specialized images". This conceptual shift allows PCExpert to leverage knowledge derived from large-scale image modality in a more direct and deeper manner, via extensively sharing the parameters with a pre-trained image encoder in a multi-way Transformer architecture. The parameter sharing strategy, combined with a novel pretext task for pre-training, i.e., transformation estimation, empowers PCExpert to outperform the state of the arts in a variety of tasks, with a remarkable reduction in the number of trainable parameters. Notably, PCExpert's performance under LINEAR fine-tuning (e.g., yielding a 90.02% overall accuracy on ScanObjectNN) has already approached the results obtained with FULL model fine-tuning (92.66%), demonstrating its effective and robust representation capability. Copyright © 2023, The Authors. All rights reserved.

关键词： Knowledge management

来源：评论

学校读者我要写书评

暂无评论

Federated data Augmentation Algorithm based on Heterogeneity Assessment 2

Federated Data Augmentation Algorithm based on Heterogeneity...

引用

2nd International Conference on Cloud Computing, Big data Application and Software engineering, CBASE 2023

作者： Hu, Bang Zhang, Mingming Liu, Kai Li, Xianhui Wang, Yongli Nanjing University of Science and Technology School of Cyber Science and Engineering Nanjing China State Grid Jiangsu Electric Power Co. Ltd. Information and Communication Branch Nanjing China State Grid Electric Power Research Institute Co. Ltd Jiangsu Ruizhong Data Co. Ltd Nanjing China Nanjing University of Science and Technology School of Computer Science and Engineering Nanjing China

ISBN: (纸本)9798350331448

In order to solve the problem of poor learning effect caused by data heterogeneity among different participants in the existing federated learning methods, this paper proposes a federated data augmentation algorithm based on heterogeneity assessment (FDA-HA) from the perspective of mitigating the effect of heterogeneity and protecting user data privacy, which mitigates the degree of data heterogeneity among participants by generating adversarial networks for data augmentation and at the same time safeguards fairness in the data augmentation process under the premise of protecting data privacy while guaranteeing fairness in the data enhancement process. Experimental results on MNIST, FashionMNIST, and Cifar10 datasets show that compared with mainstream federated learning algorithms, this algorithm improves the accuracy by 7.96% and 13.44% in data scenarios with different degrees of heterogeneity, and at the same time, it has a certain degree of fairness. © 2023 IEEE.

关键词： data privacy

来源：评论

学校读者我要写书评

暂无评论

Discriminative and robust online learning for siamese visual tracking

arXiv

引用

arXiv 2019年

作者： Zhou, Jinghao Wang, Peng Sun, Haoyang School of Computer Science and School of Automation Northwestern Polytechnical University Natl. Engineering Laboratory for Integrated Aero-Space-Ground-Ocean Big Data Application Technology

The problem of visual object tracking has traditionally been handled by variant tracking paradigms, either learning a model of the object's appearance exclusively online or matching the object with the target in an offline-trained embedding space. Despite the recent success, each method agonizes over its intrinsic constraint. The online-only approaches suffer from a lack of generalization of the model they learn thus are inferior in target regression, while the offline-only approaches (e.g., convontional siamese trackers) lack the target-specific context information thus are not discriminative enough to handle distractors, and robust enough to deformation. Therefore, we propose an online module with an attention mechanism for offline siamese networks to extract target-specific features under L2 error. We further propose a filter update strategy adaptive to treacherous background noises for discriminative learning, and a template update strategy to handle large target deformations for robust learning. Validity can be validated in the consistent improvement over three siamese baselines: SiamFC, SiamRPN++, and SiamMask. Beyond that, our model based on SiamRPN++ obtains the best results over six popular tracking benchmarks and can operate beyond real-time. Copyright © 2019, The Authors. All rights reserved.

关键词： Deformation

来源：评论

学校读者我要写书评

暂无评论

Heterogenous Multi-Source Fusion for Ship Trajectory Complement and Prediction with Sequence Modeling

Heterogenous Multi-Source Fusion for Ship Trajectory Complem...

引用

IEEE International Conference on data Science in Cyberspace (DSC)

作者： Changmeng Zheng Qi Peng Xuemiao Xu Key Laboratory of Big Data and Intelligent Robot School of Software Engineering South China University of Technology Guangzhou China School of Computer Science & Engineering South China University of Technology Guangzhou China

ISBN: (数字)9781728195582

ISBN: (纸本)9781728195599

Effectively monitoring ships and discovering abnormal ship trajectory in time is necessary for marine traffic supervision. The basic work of discovering the ship's abnormal trajectory is to predict the ship's navigation dynamically. Previous works in ship trajectory prediction are basically concern on single-source data, for example, the AIS data. These methods ignore the relations between different sources which may improve the performance of predicting ship trajectory. We propose a neural sequence model based on heterogeneous multisource fusion for ship trajectory completion and prediction. Our method makes better utilization of AIS, GPS and ARPA radar information to predict ship trajectory precisely. We construct a dataset which contains about 8 million ship trajectory samples and the experiments demonstrate that our multi-source fusion model gains promising results.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Optimized GPU implementation for dynamic programming in image data processing

Optimized GPU implementation for dynamic programming in imag...

引用

IEEE International Conference on Performance, Computing and Communications (IPCCC)

作者： Jing Ke Tomasz Bednarz Arcot Sowmya DATA 61 CSIRO University of New South Wales Sydney Australia DATA 61 CSIRO Sydney Australia School of Computer Science and Engineering University of New South Wales Sydney Australia

ISBN: (纸本)9781509052530

It is a trend now that computing power through parallelism is provided by multi-core systems or heterogeneous architectures for High Performance Computing (HPC) and scientific computing. Although many algorithms have been proposed and implemented using sequential computing, alternative parallel solutions provide more suitable and high performance solutions to the same problems. In this paper, three parallelization strategies are proposed and implemented for a dynamic programming based cloud smoothing application, using both shared memory and non-shared memory approaches. The experiments are performed on NVIDIA GeForce GT750m and Tesla K20m, two GPU accelerators of Kepler architecture. Detailed performance analysis is presented on partition granularity at block and thread levels, memory access efficiency and computational complexity. The evaluations described show high approximation of results with high efficiency in the parallel implementations, and these strategies can be adopted in similar data analysis and processing applications.

关键词： Instruction sets Dynamic programming Heuristic algorithms Smoothing methods Graphics processing units Memory management

来源：评论

学校读者我要写书评

暂无评论

Cuttlefish Algorithm-Based Deep Learning Model to Predict the Missing data in Healthcare Application

Cuttlefish Algorithm-Based Deep Learning Model to Predict th...

引用

International Conference on data Analytics and Management, ICDAM 2023

作者： Sasi Kumar, A. Rajesh Kumar, T. Balamanigandan, R. Meganathan, R. Karwa, Roshan Mahaveerakannan, R. Inurture Education Solutions Pvt. Ltd. Bangalore India Department of Cloud Technology and Data Science Institute of Engineering and Technology Srinivas University Srinivas Nagar Mukka Surathkal Mangalore India Department of Computer Science and Engineering Saveetha School of Engineering Saveetha Institute of Medical and Technical Sciences Tamil Nadu Chennai602105 India Department of Computer Science and Engineering Koneru Lakshmaiah Education Foundation AP Vaddeswaram India Department of CSE Prof Ram Meghe Institute of Technology and Research Badnera-Amravati India

ISBN: (纸本)9789819965526

Countless sensors embedded in IoT devices produce an ocean of data. The quality of IoT services depends on this information;hence, its accuracy is critical. Unfortunately, noise, collision, unreliable network connectivity, failed equipment, and manual system shutdown are all common causes of missing and partial values in IoT data. The use of predictive analytics to EMRs, which include healthcare data, inspired this effort. Predictive models based on such data are inherently flawed because they include noise, missing data, and inconsistencies between classes of interest. We argue for the need to create specialised methods of data preprocessing and categorisation since typical data mining approaches frequently provide unsatisfactory performance measurements. To build the long-term predictor in the attendance of significant nonlinearity and noise in the missing data, in this study, each of us provides a bidirectional self-attentive encoding and decoding architecture (BEDA). Prior to any further processing, the raw data is subjected to a wavelet threshold filter for denoising purposes. The weights are optimised via a mapping-based cuttlefish optimisation technique, and the bidirectional long short-term-memory is chosen as the basic device for isolating temporal and serial features (MCFA). The prediction performance is then enhanced by adding the multi-head self-attention device into the encoder–decoder architecture. It demonstrates that the projected technique yields quick, accurate, and resilient classification consequences by comparing them to results from other models on public data sets with unbalanced classes and missing values in health-related applications. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd 2023.

关键词： data mining

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：