检索结果-内蒙古大学图书馆

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

作者： Kun Yang Jun Lu College of Computer Science and Technology Heilongjiang University Harbin China Jiaxiang Industrial Technology Research Institute of HLJU Jining China Key Laboratory of Database and Parallel Computing of Heilongjiang Province Harbin China

The proposed method in this paper proposes an end-to-end unsupervised semantic segmentation architecture DMSA based on four loss functions. The framework uses Atrous Spatial Pyramid Pooling (ASPP) module to enhance feature extraction. At the same time, a dynamic dilation strategy is designed to better capture multi-scale context information. Secondly, a Pixel-Adaptive Refinement (PAR) module is introduced, which can adaptively refine the initial pseudo labels after feature fusion to obtain high quality pseudo labels. Experiments show that the proposed DSMA framework is superior to the existing methods on the saliency dataset. On the COCO 80 dataset, the MIoU is improved by 2.0, and the accuracy is improved by 5.39. On the Pascal VOC 2012 Augmented dataset, the MIoU is improved by 4.9, and the accuracy is improved by 3.4. In addition, the convergence speed of the model is also greatly improved after the introduction of the PAR module.

关键词： Training Representation learning Adaptation models Semantic segmentation Semantics Signal processing Transformers

来源：评论

学校读者我要写书评

暂无评论

Design of Radio-over-Fiber Distributed Antenna System and Application of Wireless Positioning

Design of Radio-over-Fiber Distributed Antenna System and Ap...

引用

IEEE Information Technology, Networking, Electronic and Automation Control Conference

作者： Hao Yun Wei Ji Rui Yin Key Laboratory of Database and Parallel Computing of Heilongjiang Province Laboratory of Intelligent Communications and Multimedia Information Processing ShanDong University Qingdao China School of Information Science and Engineering ShanDong University Qingdao China

ISBN: (数字)9781728143903

ISBN: (纸本)9781728143910

In this paper, a radio-over-fiber distributed antenna system (ROF-DAS) and application of wireless positioning are proposed. The modified Medium Access Control (MAC) frame, optical switch array, ROF network based on star topology and Reflective Semiconductor Optical Amplifier (RSOA) are employed to design the system. Based on the structure, wireless positioning could be achieved by Time Division Multiplexing-Time Difference of Arrival (TDM-TDOA) technology combined with Chan assisted Taylor algorithm. The quality of signal transmission is evaluated by Q Factor, and the positioning accuracy is performed in terms of the antennas amount using the positioning algorithm.

关键词：

来源：评论

学校读者我要写书评

暂无评论

User Local and Global Interest Depth Interaction for Recommendation Algorithm

User Local and Global Interest Depth Interaction for Recomme...

引用

International Conference on Information Science and Control Engineering (ICISCE)

作者： Xufeng Zhang Yan Yang School of Computer Science Technology Heilongiiang University Harbin China Key Laboratory of Database and Parallel Computing Harbin Heilongiiang Province China

Using auxiliary information such as social networks or item attributes to improve the performance of recommender systems is a key issue in the research of recommender systems. It plays a very important role in alleviating the interactive sparsity of collaborative filtering methods. In response to this problem, a recommendation algorithm that user local and global interest depth interaction is proposed, which analyzes the user’s implicit preferences from two perspectives: based on the user’s historical behavior records, using a fully connected neural network to extract the user’s local interests; based on the user’s historical behavior items, through the association between entities in the knowledge graph, mines the entities associated with the user’s historical behavior, forming a multi-layer global interest layer to generalize user interests. The user’s local interest vector activates the item entity in the global interest layer to generate the user’s global interest vector. Finally, the deep neural network is used to complete the feature interaction of users’ local interests, global interests and recommended items. The user’s local and global interests accurately describe user preferences and improve the accuracy of click-through rate prediction. Experiments show that the recommendation performance of the algorithm on the two public data sets is better than the comparison baseline and the classic recommendation algorithm.

关键词： Knowledge engineering Deep learning Information science Control engineering Social networking (online) Neural networks Feature extraction

来源：评论

学校读者我要写书评

暂无评论

NKMH: A Neural Efficient Recommendation Based on Neighborhood key Information Aggregation of Modified Hawkes

NKMH: A Neural Efficient Recommendation Based on Neighborhoo...

引用

IEEE Annual International Computer Software and Applications Conference (COMPSAC)

作者： Xin Xu Nan Wang Hu Jin Yang Liu Kun Li Key Laboratory of Database and Parallel Computing of Heilongjiang Province School of Computer Science and Technology Heilongjiang University Harbin China Postdoctoral Program of Heilongjiang Hengxun Technology Co. Ltd School of Computer Science and Technology Heilongjiang University Harbin China

ISBN: (数字)9781665488105

ISBN: (纸本)9781665488112

The rapid development of neural networks has con-tributed to the increasing maturity of recommendation systems. However, deep neural networks have poor interpretability for models and do not show strong advantages for sparse data and noisy data. Recently, Hawkes process has become more and more focused for its good interpretability with probabilistic models. Based on this, we proposes A Neural Efficient Recommendation Model Based on Neighborhood key Information Aggregation of Modified Hawkes(NKMH). The model utilizes a neural network and designs three modules to jointly fit the modified Hawkes process. It not only inherits the high interpretability of Hawkes, but also effectively solves the problem of poor prediction ability of the Hawkes process. Besides, we present a novel key information search strategy(KISS), which can effectively remove the noise in a session and alleviate the sparsity of the data to some extent. Extensive experiments on two datasets show that the NKMH model outperforms many current popular models.

关键词： Deep learning Conferences Neural networks Predictive models Search problems Probabilistic logic Software

来源：评论

学校读者我要写书评

暂无评论

Faster BERT with Double Loss Function Based on Batch Data Preprocessing

Faster BERT with Double Loss Function Based on Batch Data Pr...

引用

作者： Pengqi Duan Yudong Zheng Jun Lu College of Computer Science and Technology Heilongiiang University Heilongjiang UniversityHLJU College of Computer Science and Technology Heilongjiang University Heilongjiang UniversityHLJU College of Computer Science and Technology Heilongiiang University Key Laboratory of Database and Parallel Computing of Heilongjiang Province Heilongjiang University Heilongjiang UniversityHLJU

Even though pre-trained language models like BERT and XLNet have produced significant consequences on a variety of tasks of natural language processing,they are difficult to deploy in practical applications due to their large number of parameters,and huge calculation and memory *** order to improve the training speed and inference speed of the pretrained model,the FasterBERT model is proposed in this *** model combines batch data preprocessing method with knowledge *** increases the model to increase the model speed and maintains the original ***,double loss function is incorporated during the knowledge distillation process in this paper,so that the model can better learn the knowledge of the teacher model and the knowledge of the real *** accuracy of the model is *** effectiveness of the FasterBERT model is demonstrated on 12 datasets in this *** speed of the FasterBERT is 1.1-2.2 times higher than FastBERT and 2-43 times higher than *** addition,the accuracy is up to 0.6% higher than that of FastBERT,and the accuracy is better than that of the BERT model on three *** experiments show that the FasterBERT model proposed in this paper performs well on six Chinese datasets and six English *** speed is greatly improved while the accuracy of the task also has a significant effect.

关键词： pre-trained language model FasterBert model double loss function batch data processing knowledge distillation

来源：评论

学校读者我要写书评

暂无评论

A Sleep Stage Classification Method via Combination of Time and Frequency Domain Features based on Single-Channel EEG

A Sleep Stage Classification Method via Combination of Time ...

引用

IEEE International Conference on Big Data and Cloud computing (BdCloud)

作者： Caihong Zhao Wenpeng Neng College of Computer Science and Technology College of Electronic Engineering Heilongjiang University Heilongjiang Key Laboratory of Database and Parallel Computing Harbin China College of Computer Science and Technology Heilongjiang University Harbin China

Sleep staging is an important method to diagnose and treat insomnia, sleep apnea, and other sleep disorders. Compared with the multi-channel automatic sleep staging system, the single-channel EEG signal contains less information, and the traditional single analysis domain feature parameter extraction algorithm cannot meet the requirement of sleep stage classification accuracy. To solve this problem, we propose an automatic sleep staging method based on the combination of time-domain and frequency-domain features based on single-channel EEG signals. Empirical mode decomposition is used to decompose EEG signals in the time domain to obtain the decomposed signals at different time scales. Multiple local features are extracted from each decomposed signal. The frequency-domain features of EEG signals are obtained by using the frequency domain decomposition of EEG signals in various rhythms. The time-domain and frequency-domain decomposition features are combined into eigenvectors and selected for sleep staging. The experimental results show that the sleep staging method proposed in this paper with time-frequency domain features of single-channel EEG signals can approach the accuracy of sleep staging of multi-channel signals on the same data set, and superior to the sleep staging method with the same single-channel EEG signals.

关键词： Deep learning Time-frequency analysis Empirical mode decomposition Sleep Frequency-domain analysis Feature extraction Electroencephalography

来源：评论

学校读者我要写书评

暂无评论

An implementation of the QR iterations for finding eigenvalues of matrices with CUDA on GPU

An implementation of the QR iterations for finding eigenvalu...

引用

International Conference on Mechatronic Science, Electric Engineering and Computer

作者： Meirui Meihui Weiping Zhang Tao Wang Ning Tian Jinbao Li Longjiang Guo School of Computer Science and Technology Heilongjiang University Harbin 150080 China Harbin Research Institute of Electrical Instrumentation 150028 China School of Computer Science and Technology Heilongjiang University Harbin China Key Laboratory of Database and Parallel Computing of Heilongjiang Province Harbin China

ISBN: (纸本)9781479925667

Based on the principle of QR loop iterations, this paper implements a parallel algorithm based on the hardware of GPU (Graphic Process Unit) by using routines from CUDA (Computer Unified Device Architecture) to find the eigenvalues of general matrices. CPU and GPU computing card form a client-server computing framework. Here, CPU can be regarded as a client and GPU card is considered as a computing server. For the experiment environment, this paper chooses a GPU card with the model of NVIDIA GeForce GTX460 as a server side and a CPU with the model of Intel Core i5-760 quad-core as a client side. Win7 64-bit is selected as the operating system. The parallel implementation consists of two parts:PA H and PA QR. PA H is a procedure that transforms a matrix A into the Hessenberg matrix B. PA QR is the actual parallel algorithm of the QR iterations that is imposed on the Hessenberg matrix B for finding eigenvalues of the general matrix A. The speedup ratio of the proposed algorithm is jarless when the number of iterations becomes greater. The experimental results show that the parallel implementation with CUDA on GPU only makes use of less running time than traditional sequential algorithms. The speedup ratio of PA H is between 1.79 and 7.81. The speedup ratio of PA QR is between 3.24 and 118.9. Especially, when the order of general matrix equals 8192, the amount of iterations becomes 10000. The speedup ratio of the PA-H and the PA QR can run up to 7.81 and 118.9 respectively.

关键词： Graphics processing units Eigenvalues and eigenfunctions Hardware Kernel Instruction sets

来源：评论

学校读者我要写书评

暂无评论

Power adjustable cross-layer network protocol in dense wireless sensor networks

引用

Ruan Jian Xue Bao/Journal of Software 2016年 27卷 11-20页

作者： Ni, Lin-Yu Li, Jin-Bao School of Computer Science and Technology Heilongjiang University Harbin150080 China Key Laboratory of Database and Parallel Computing of Heilongjiang Province Harbin150080 China

A cross layer network strategy with power regulation is proposed for the wireless sensor networks with dense nodes. Under this strategy, nodes use different transmission power for data transmission and so that each node has a predetermined number of its neighbor nodes, or node degree. It thus guarantees the network connectivity, and can effectively reduce the competition conflict between nodes. The nodes in the network are divided according to the hierarchy. The upper node with the minimum transmission power is selected as the father node to construct the routing tree, and the candidate father node sets for the nodes are calculated. In the case of the failure of the father node, candidate father nodes are selected for transmission to solve the problem. The experimental results show that this strategy can improve the network throughput and the rate of data transmission. © Copyright 2016, Institute of Software, the Chinese Academy of Sciences. All rights reserved.

关键词： Throughput

来源：评论

学校读者我要写书评

暂无评论

Study on distributed channel allocation and routing forwarding strategy in Dual-Radio sensor networks

引用

Tongxin Xuebao/Journal on Communications 2010年第9 A期31卷 81-89页

作者： Fan, Bin Li, Jin-Bao Guo, Long-Jiang Liu, Xu Ren, Qian-Qian School of Computer Science and Technology Heilongjiang University Harbin 150080 China Key Laboratory of Database and Parallel Computing of Heilongjiang Province Harbin 150080 China

Considering the limited orthogonal channels in wireless sensor networks (WSN), a distributed channel allocation algorithm (DCA) for Dual-Radio WSN was presented. Based on DCA, a routing forwarding strategy (RFS) was proposed as well. DCA avoids conflictions within a path and reduces conflictions among different paths with a small number of channels. Furthermore theoretically prove that when K≥Δ+1, the upper bound of the network throughput is |M|+1, where K is the number of orthogonal channels in WSN, Δ is the largest degree of the network topology graph, and |M| is the number of non-leaf nodes in the network routing tree. Nodes switch channels timely in RFS, and then further dropping the conflict between the paths, effectively transmit data in parallel. Simulation results indicate that DCA and RFS can reduce the latency of data forwarding and the average energy consumption of nodes, as well as increase the throughput significantly.

关键词： Energy utilization

来源：评论

学校读者我要写书评

暂无评论

Energy-efficient algorithm for mobile target prediction and tracking in wireless sensor networks

引用

Tongxin Xuebao/Journal on Communications 2010年第9 A期31卷 135-143页

作者： Liu, Xu Li, Jin-Bao Guo, Long-Jiang Fan, Bin Zhu, Jing-Hua School of Computer Science and Technology Heilongjiang University Harbin 150080 China Key Laboratory of Database and Parallel Computing of Heilongjiang Province Harbin 150080 China

Aiming at the problem of target tracking in wireless sensor networks, an energy efficient algorithm for mobile targets prediction and tracking was proposed. Sleep scheduling mechanism was used to reduce the cost, as well as to guarantee the real-time tracking. The algorithm predicts the trajectory of the target with the Markov chain theory, and the sleeping nodes in the predicted area would be waked up to monitor the targets. In order to predict the target accurately, a location algorithm based on distance vector was triggered to estimate its position. The sensors sleep initiatively to save energy when they did not have the sensing task. Simulation and testbed experimental results indicate that the algorithms can accurately describe the target trajectory, and efficiently reduce the energy consumption.

关键词： Target tracking

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：