检索结果-内蒙古大学图书馆

arXiv 2020年

作者： Shen, Lei Feng, Yang Key Laboratory of Intelligent Information Processing Institute of Computing Technology Chinese Academy of Sciences Beijing China University of Chinese Academy of Sciences Beijing China

Emotion-controllable response generation is an attractive and valuable task that aims to make open-domain conversations more empathetic and engaging. Existing methods mainly enhance the emotion expression by adding regularization terms to standard cross-entropy loss and thus influence the training process. However, due to the lack of further consideration of content consistency, the common problem of response generation tasks, safe response, is intensified. Besides, query emotions that can help model the relationship between query and response are simply ignored in previous models, which would further hurt the coherence. To alleviate these problems, we propose a novel framework named Curriculum Dual Learning (CDL) which extends the emotion-controllable response generation to a dual task to generate emotional responses and emotional queries alternatively. CDL utilizes two rewards focusing on emotion and content to improve the duality. Additionally, it applies curriculum learning to gradually generate high-quality responses based on the difficulties of expressing various emotions. Experimental results show that CDL significantly outperforms the baselines in terms of coherence, diversity, and relation to emotion factors. Copyright © 2020, The Authors. All rights reserved.

关键词： Curricula

来源：评论

学校读者我要写书评

暂无评论

Optical Character recognition Using Optimized Convolutional Networks*

Optical Character Recognition Using Optimized Convolutional ...

引用

International Conference on Fog and Mobile Edge Computing (FMEC)

作者： Anum Nawaz Muhammad Irfan Tomi Westerlund Shanghai Key Laboratory of Intelligent Information Processing School of Computer ScienceFudan University Shanghai China Turku Intelligent and Embedded Robotic Systems Lab Faculty of Technology University of Turku Turku Finland

Optical Character recognition (OCR) has been a prominent area of research in pattern recognition for several decades, owing to its broad application potential in smart living. To improve offline OCR on mobile devices with limited computing resource, we have optimized Convolutional Neural Networks (CNNs) to efficiently detect text using minimal resources. To achieve this, we employed two distinct pretrained CNN models, namely AlexNet and Inception-V3, for feature extraction. Leveraging these models' unique characteristics and capabilities to extract diverse features, we aimed to enhance the classifier's accuracy. This, in turn, facilitates the development of an efficient edge-device application for faster and higher-quality OCR. Experimental results demonstrate that our proposed optimized algorithm outperforms existing CNN-based methods in the field of OCR, particularly in the categorization and detection of handwritten digits and character recognition. The conducted research yielded impressive accuracy results, with up to 97% accuracy on the MNIST dataset and 95.5% accuracy on the NIST dataset.

关键词：

来源：评论

学校读者我要写书评

暂无评论

High Depth Resolution Wavelength Scanning Interferometry with Narrow Scanning Bandwidth Via Sinewaves Separation

SSRN

引用

SSRN 2022年

作者： Tan, Ji He, Zhaoshui Bai, Yulei Dong, Bo Xie, Shengli School of Automation Guangdong University of Technology Guangzhou510006 China Key Laboratory of Intelligent Information Processing and System Integration of IoT Ministry of Education Guangzhou510006 China Key Laboratory of Intelligent Detection The Internet of Things in Manufacturing Ministry of Education Guangzhou510006 China Guangdong Key Laboratory of IoT Information Technology Guangzhou510006 China

Wavelength scanning interferometry (WSI) is a promising tomographic imaging technique. However, the depth resolution is fundamentally limited by a narrow wavelength scanning bandwidth, which brings challenges to the frequency extraction for depth information. In this work, we proposed a parameter estimation (PE) paradigm via sinewaves separation, which allows high depth resolution tomographic measurement with narrow-band sources. To obtain accurate depth information, a sinewaves separation model is designed to estimate the frequency of the overlapped signals and it can achieve higher depth resolution without any prior knowledge even under limited scanning bandwidth. Moreover, the estimated frequency is used for phase retrieval and the multi-layer 3D profile measurement can be further implemented. The proposed method has been applied to measure the multi-layer acrylic plastic and USAF resolution board. The experimental results show that the performance of the proposed method is comparable to DFT even if reducing about 60% scanning bandwidth, i.e., the proposed method can improve approximately 3 folds of the depth resolution. © 2022, The Authors. All rights reserved.

关键词： Bandwidth

来源：评论

学校读者我要写书评

暂无评论

A rough RBF neural networks optimized by the genetic algorithm

引用

Advances in information Sciences and Service Sciences 2011年第7期3卷 332-339页

作者： Ding, Shifei Ma, Gang Xu, Xinzheng School of Computer Science and Technology China University of Mining and Technology Xuzhou China Key Laboratory of Intelligent Information Processing Institute of Computing Technology Chinese Academy of Science Beijing China

The large-scale data parallelism processing is an inherent characteristic of artificial neural networks, but the networks bring the efficiency problems of data processing. As one of the artificial neural networks, Radial Basis Function (RBF) neural networks have the same problem. Therefore, how to reduce the scale of data to improve the efficiency of data processing has been a hot issue among the artificial intelligence scholars. Based on the traditional RBF neural networks, this paper puts forward a method which determines the important degree of the sample attributes based on knowledge entropy of Rough set by analyzing the relationship between the knowledge entropy and the weight of the sample attributes, and assesses the importance of the sample attributes between the input layer and the hidden layer, namely the attribution reduction, so as to achieve reduce the scale of data processing. The ultimate aim of training RBF neural networks is to seek a set of suitable networks parameters which makes the sample output error achieve the minimum or required accuracy, while Genetic Algorithm (GA) has the properties of finding out the optimal solution through multiplepoint random search in the solution space, so Genetic Algorithm is used to optimize the centers, the widths and the weights between the hidden layer and the output layer of RBF neural networks in training the networks. Finally, a model about A Rough RBF Neural Networks Optimized by the Genetic Algorithm (GA-RS-RBF) is proposed in this paper. The simulation results show that the rough RBF neural network optimized by the Genetic Algorithm is better than the traditional RBF neural networks in classification about Iris datasets.

关键词： Genetic algorithms

来源：评论

学校读者我要写书评

暂无评论

A Profile Matching Scheme based on Private Set Intersection for Cyber-Physical-Social Systems

A Profile Matching Scheme based on Private Set Intersection ...

引用

IEEE Conference on Dependable and Secure Computing

作者： Yalian Qian Xueya Xia Jian Shen School of Computer and Software Nanjing university of information science and technology Nanjing China Fujian Provincial Key Laboratory of Networking Computing and Intelligent Information Processing Fuzhou University Fujian China

ISBN: (数字)9781728175348

ISBN: (纸本)9781728175355

The cyber-physical-social system (CPSS) is a three-layer system framework that combines the human society on the basis of the cyber-physical system (CPS), so that the human society, the cyber world and the physical world are interconnected. In the CPSS, similar profile attributes are matched to socialize and ultimately achieve the purpose of information sharing. However, some personal information may be included in the profile attributes, thus the users' privacy cannot be protected during the process. To meet this challenge, a privacy-preserving profile matching scheme based on private set intersection is proposed in this paper. Multi-tag is utilized to partition the dataset of users to achieve fine-grained profile matching. In addition, the privacy of users is protected by re-encryption technique. Security analysis shows that our scheme is secure against the semi-honest adversary and theoretical analysis of the experiment shows that that the scheme is efficient for profile matching in the CPSS.

关键词： Privacy Conferences Resists Cyber-physical systems Performance analysis information management Security

来源：评论

学校读者我要写书评

暂无评论

A hybrid deep learning CNN-ELM model and its application in handwritten numeral recognition

引用

Journal of Computational information Systems 2015年第7期11卷 2673-2680页

作者： Guo, Lili Ding, Shifei School of Computer Science and Technology China University of Mining and Technology Xuzhou China Key Laboratory of Intelligent Information Processing Institute of Computing Technology Chinese Academy of Sciences Beijing China

Convolutional Neural Network (CNN) is a typical algorithm structure of deep learning and it has applied in image recognition field widely. Based on CNN, this paper puts forward a novel hybrid deep learning model CNN-ELM. In this model, CNN is in charge of feature extraction and Extreme Learning Machine (ELM) performs as a classifier to complete the final classification so as to integrate the synergy of two classifiers. We have took experiments on MNIST digit database. Comparisons with CNN and ELM on the same database, CNN-ELM achieved an error rate of 0.67%, which indicate that CNN-ELM has achieved better generalization performance. ©, 2015, Binary information Press. All right reserved.

关键词： Character recognition

来源：评论

学校读者我要写书评

暂无评论

RESEARCH ON CHINESE FAQ QUESTION ANSWERING SYSTEM IN RESTRICTED DOMAIN

RESEARCH ON CHINESE FAQ QUESTION ANSWERING SYSTEM IN RESTRIC...

引用

2007 International Conference on Machine Learning and Cybernetics(IEEE第六届机器学习与控制论国际会议)

作者： ZHENG-TAO YU YAN-XIA QIU JIN-HUI DENG LU HAN XIANG-YAN MENG CUN-LI MAO The School of Information Engineering and Automation Kunming University of Science and Technology The School of Information Engineering and Automation Kunming University of Science and Technology The Institute of Intelligent Information Processing Computer Technology Application Key Laboratory

FAQ(Frequently-asked Question) is a good question and answer model to realize business advisory system in restricted domain.A FAQ question answering system model is presented in this *** the help of the idea of ontology, a knowledge base is constructed in the *** the help of language KDML (Knowledge Database Mark-up Language) of HowNet,the domain ontology and the relationship of it are defined and described, and the fusion of domain knowledge base (Domain HowNet) and common knowledge base(HowNet) is *** this basis, a question similarity calculation method, which makes use of the characteristics of the domain question and combines lexical relationship, syntactic interdependent relationship and the semantic relationship of domains among question sentences, is *** based on the question similarity calculation, retrieval of related question from the candidate question set and extraction of answers can be implemented with this *** result of Yunnan tourism question-answer model experiment shows that this method is feasible and effective.

关键词： Restricted domain question answering system Frequently-asked question Domain ontology Question similarity

来源：评论

学校读者我要写书评

暂无评论

CPR-Net: A Novel Joint Learning Network for Pulmonary Nodule Evaluation

CPR-Net: A Novel Joint Learning Network for Pulmonary Nodule...

引用

IEEE International Conference on Systems, Man and Cybernetics

作者： Gong Chen Jun Liu Chen-Qian Li Wuhan University of Science and Technology Wuhan China Hubei Province Key Laboratory of Intelligent Information Processing and Real-Time Industrial System Wuhan China

Lung cancer is the most common malignant tumor worldwide, with high mortality rates. Pulmonary nodule is a common manifestation of lung cancer. Accurate segmentation and detection of pulmonary nodules from CT scans are essential for proper assessment of patient prognosis. However, this task remains challenging due to various factors, such as class imbalance and the need for detailed characterization of lung nodule segmentation. To address these issues, we propose a novel network, CPR-Net (Convolutional Pyramid Residuals-Net), that combines segmentation and benign-malignant classification for CT images of lung nodules. Our approach utilizes a PBMM module to expand the perception field and enhance the representation capability of the model, allowing it to learn more detailed information. We evaluate the classification performance using accuracy, precision, recall, F1 score, AUC, and the segmentation performance using the Dice coefficient. Experiments on both publicly available datasets and self-constructed datasets demonstrate that our proposed method in this paper outperforms other methods in terms of classification and segmentation performance under limited labeled data conditions. Moreover, Our results suggest that this model has great potential for improving lung nodule diagnosis in radiology for heterogeneous intranodal images.

关键词：

来源：评论

学校读者我要写书评

暂无评论

An Improved RANSAC Algorithm Based on Similar Structure Constraints

An Improved RANSAC Algorithm Based on Similar Structure Cons...

引用

International Conference on Robots & intelligent System (ICRIS)

作者： Wenqiu Zhu Wenjing Sun Yexiang Wang Shaolin Liu Keke Xu School of Computer and Communication Hunan University of Technology Zhuzhou China Key Laboratory of Intelligent Information Perception and Processing Technology Zhuzhou Hunan Province China

ISBN: (纸本)9781509041565

An improved RANSAC algorithm based on structural similarity was proposed to improve the speed and accuracy of traditional RANSAC (Random Sample Consensus) algorithm and to reduce iterations and runtime. Firstly, BRISK (Binary Robust Invariant Scalable keypoints) algorithm was used to extract and describe feature points. The initial match set is obtained by hamming distance feature matching. Then, mismatches are eliminated by similar structure constraints. Finally, the new match set is taken as the input of RANSAC to calculate the transformation matrix. The algorithm can obtain the transformation model quickly because it has purified matching points after the initial matching. Experiments show that the number of iterations and runtime of this algorithm are obviously less than the number of iterations and runtime of the traditional algorithm. So the proposed method outperforms traditional RANSAC (Random Sample Consensus) algorithm significantly both in iterations and speed.

关键词： Robots intelligent systems

来源：评论

学校读者我要写书评

暂无评论

A method of laser point temperature detection based on CNN

A method of laser point temperature detection based on CNN

引用

Chinese Automation Congress (CAC)

作者： Ruidong Xie Haichuan Yang Yingmin Yi Qimeng Shi keying Wang Jiabin Huang Key Lab of Manufacturing Equipment of Shaanxi Province Xi’an University of Technology Xi’an China Shaanxi Key Laboratory of Complex System Control and Intelligent Information Processing Xi’an University of Technology Xi’an China

ISBN: (数字)9781728176871

ISBN: (纸本)9781728176888

There has been a direct relationship between the temperature of the laser point and the quality of the casting in the process of the 3D printing. In the paper, a method based on convolutional neural network (CNN) was proposed to estimate the temperature of the laser point. The collected temperature data used were trained by the deep-learning method. A new structure of the model was proposed on the part of the CNN model, which improved from original LeNet. The process of the prediction for the testing set was carried out through the new model. The unknown temperature in the testing set can be estimated. The experimental result illustrates that the proposed method is satisfactory.

关键词： Laser modes Temperature measurement Training Temperature distribution Surface emitting lasers Testing Measurement by laser beam

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：