检索结果-内蒙古大学图书馆

arXiv 2018年

作者： Shao, Kun Zhu, Yuanheng Zhao, Dongbin State Key Laboratory of Management and Control for Complex Systems Institute of Automation Chinese Academy of Sciences Beijing100190 University of Chinese Academy of Sciences Beijing

Real-time strategy games have been an important field of game artificial intelligence in recent years. This paper presents a reinforcement learning and curriculum transfer learning method to control multiple units in StarCraft micromanagement. We define an efficient state representation, which breaks down the complexity caused by the large state space in the game environment. Then a parameter sharing multi-agent gradientdescent Sarsa(a;) (PS-MAGDS) algorithm is proposed to train the units. The learning policy is shared among our units to encourage cooperative behaviors. We use a neural network as a function approximator to estimate the action-value function, and propose a reward function to help units balance their move and attack. In addition, a transfer learning method is used to extend our model to more difficult scenarios, which accelerates the training process and improves the learning performance. In small scale scenarios, our units successfully learn to combat and defeat the built-in AI with 100% win rates. In large scale scenarios, curriculum transfer learning method is used to progressively train a group of units, and shows superior performance over some baseline methods in target scenarios. With reinforcement learning and curriculum transfer learning, our units are able to learn appropriate strategies in StarCraft micromanagement scenarios. Copyright © 2018, The Authors. All rights reserved.

关键词： Reinforcement learning

来源：评论

学校读者我要写书评

暂无评论

Determination of Polynomial Degree in the Regression of Drug Combinations

引用

IEEE/CAA Journal of Automatica Sinica 2017年第1期4卷 41-47页

作者： Boqian Wang Xianting Ding Fei-Yue Wang School of Biomedical Engineering Shanghai Jiao Tong University IEEE State Key Laboratory of Management and Control for Complex Systems Institute of Automation Chinese Academy of Science (SKL-MCCS CASIA) Research Center of Computational Experiments and Parallel Systems National University of Defense Technology

Studies on drug combinations are becoming more and more popular in the past few decades, with the development of computer and algorithms. One of the most common methods in optimizing drug combinations is regression of a polynomial model based on certain number of experimental observations. In this paper, we study how to determine the degree of polynomials in different circumstances of drug combination optimization. Using cross-validation, we have found that in most cases, a high degree results in failures of accurate prediction, named overfitting. An anti-noise test has also revealed that polynomial model with high degree tends to be less resistant to random errors in the observations. © 2014 Chinese Association of Automation.

关键词： Acoustics Random errors Regression analysis

来源：评论

学校读者我要写书评

暂无评论

Recognize User Intents in Online Interactions from Massive Social Media Data 2

Recognize User Intents in Online Interactions from Massive S...

引用

2017 IEEE 2nd International Conference on Big Data Analysis（ICBDA 2017）

作者： Chenxi Cui Wenji Mao State Key Laboratory of Management and Control for Complex Systems Institute of Automation Chinese Academy of Sciences School of Computer and Control Engineering University of Chinese Academy of Sciences

ISBN: (纸本)9781509036202

Online interactions,especially user generated contents on social events,reveal a variety of communicative purposes ranging from expressing feelings to proposing *** intents in users' online interactive behavior from massive social media data can effectively identify users' motives and intents behind communication and provide important information to aid monitoring,analysis and decision making for a variety of ***,user intents recognition from online communication is inherently challenging due to the ambiguity in semantic processing and diversity of syntax ***,the massive online data are usually unlabeled,which greatly hinders the usage of typical machine learning based methods that can automate the recognition *** this paper,we tackle this problem by proposing a Speech Act Theory guided classification scheme,which regards online communication as performative actions of users and classifies user utterances according to their pragmatic *** the basis of this,we construct a dictionary of performative words,expand it using external knowledge sources and refine it by word embedding and similarity *** then use this dictionary to automatically label the online textual data with *** a large amount of the labeled data,we train feature based classifiers to identify user intents in their online *** experimental study using a microblog dataset on social events from Sina Weibo shows the effectiveness of our proposed method.

关键词： user intent recognition speech act theory online communication social media big data

来源：评论

学校读者我要写书评

暂无评论

A Residual Merged Neutral Network for Multimodal Sentiment Analysis 2

A Residual Merged Neutral Network for Multimodal Sentiment A...

引用

2017 IEEE 2nd International Conference on Big Data Analysis（ICBDA 2017）

作者： Nan Xu Wenji Mao State Key Laboratory of Management and Control for Complex Systems Institute of Automation Chinese Academy of Sciences School of Computer and Control Engineering University of Chinese Academy of Sciences

ISBN: (纸本)9781509036202

With the continuous development of social networking sites,the volume of social media data has exploded and the user-generated content is becoming more and more *** a result,the modality of massive social media data is no longer confined to the single text *** brings new challenges to social media analytics in general and its examplar field such as sentiment analysis in *** sentiment analysis has become an increasingly important research topic in recent years,especially in the context of social media big *** of the previous work only focuses on single modality content such as text,image or ***,as the traditional sentiment analysis methods often lack the support of scalable deep models,this hinders their usage in processing large amount of online *** overcome the limitations in the previous work,in this paper,we propose an end-to-end framework for multimodal sentiment analysis based on deep neural *** propose a Merged Neural Network(MNN) model that utilizes CNNs to extract representations of text and image *** fuse the multimodal features,we introduce the residual model and propose two combined merged strategies,namely the EarlyRMNN(*** Residual MNN) and Late-RMNN(*** Residual MNN),to get deeper and more discriminative features than the previous *** experiments on two public available datasets demonstrate the effectiveness of our models for multimodal sentiment analysis in comparison with the related methods.

关键词： multimodal sentiment analysis convolutional neural network residual model feature fusion social media big data

来源：评论

学校读者我要写书评

暂无评论

Efficient Deployment of Patrols to Catch Arsonists

Efficient Deployment of Patrols to Catch Arsonists

引用

Chinese Automation Congress (CAC)

作者： Jinpeng Han Fei-Yue Wang Yisheng Lv Fenghua Zhu Yilun Lin Xingyuan Dai Li Li School of Software Engineering Xian Jiaotong University Xian China The State Key Laboratory for Management and Control of Complex Systems Chinese Academy of Sciences Beijing Qingdao Academy of Intelligent Industries Qingdao Shandong The State Key Laboratory for Management and Control of Complex Systems Institute of Automation Chinese Academy of Science Beijing China Department of Automation Tsinghua University Beijing China

Forest fires are serious disasters that endanger the safety and property of residents. To deal with the arson-caused forest fires, we use security game theory to generate optimal patrol strategy under limited patrolling resources. The existing security game algorithms focus on a single behavior model to prevent attacks, while our work is to incorporate the secondary damage caused by the attack behavior into the optimization object. For forest fires, the security game algorithm only expects to catch arsonists in the forest, which is unreasonable in our opinion. Our algorithm incorporates the spread of fires into patrolling path planning so that defenders avoid being swallowed by fire via capturing the spread of it.

关键词： Games Forestry Security Game theory Mathematical model Automation control systems

来源：评论

学校读者我要写书评

暂无评论

Catching Dynamic Heterogeneous User Data for Identity Linkage Learning

Catching Dynamic Heterogeneous User Data for Identity Linkag...

引用

International Joint Conference on Neural Networks

作者： Fan Lei Qiudan Li Song Sun Lei Wang Daniel Dajun Zeng The State Key Laboratory of Management and Control for Complex Systems Institute of Automation Chinese Academy of Sciences Beijing China University of Chinese Academy of Sciences Beijing China

Benefitting from the development of social platforms, more and more users tend to register multiple accounts on different social networks. Linking user identities across multiple online social networks based on user behavior patterns is considerable for network supervision and information tracking. However, a user's online behavior in a social network is dynamic. The user profile may be changed due to some specific reasons such as user migration or job changes. Thus, catching the dynamics of evolutionary user data and collecting the latest user features are important and challenging issues in the area of user identity linkage. Inspired by deep learning models such as word2vec and DeepWalk, this paper proposes an integrated framework to catch the dynamic user data by supplementing vacant features and updating outdated features in data sources. The framework firstly represents all textual and structural user data into low-dimensional latent spaces by utilizing word2vec and DeepWalk, then, integrates different user features and predicts vacant data fields based on late fusion approach and cosine similarity computation. We then explore and evaluate the application of our proposed method in a user identity mapping task. The results proved that our framework can successfully catch the dynamic user data and enhance the performance of identity linkage models by supplementing and updating data sources advance with the times.

关键词： Couplings Data models Twitter Data mining Machine learning Task analysis

来源：评论

学校读者我要写书评

暂无评论

An image processing method for green apple lesion detection in natural environment based on GA-BPNN and SVM

An image processing method for green apple lesion detection ...

引用

IEEE International Conference on Mechatronics and Automation

作者： Yunong Tian En Li Lei Yang Zize Liang The University of Chinese Academy of Sciences Beijing China The State Key Laboratory of Management and Control for Complex Systems Institute of Automation Chinese Academy of Sciences Beijing China

The technologies of information processing, especially the image processing, are important technical means to realize intelligent management and control of modern orchard. In order to realize the intelligentization of orchard production, the orchard image database, especially the lesion image database needs to be set up. Because of the uneven illumination and the complex background under the natural light condition, traditional image segmentation methods cannot solve the adaptive threshold issue during the lesion image processing of green apples. This paper proposes an image processing method based on a BP neural network updated by genetic algorithm (GA-BPNN) and support vector machine (SVM) to realize lesion image processing of green apples in orchard. With this method, apple images can be processed in batch and the lesion image database can be consummated automatically. Furthermore, the recognition of the diseased apple is realized. The experimental results show that the proposed method can obtain green apple segmentation and lesion detection with good efficiency and robustness in complex natural orchard environment.

关键词： Lesions Image segmentation Diseases Training Green products Support vector machines

来源：评论

学校读者我要写书评

暂无评论

Learning Insulators Segmentation from Synthetic Samples

Learning Insulators Segmentation from Synthetic Samples

引用

International Joint Conference on Neural Networks

作者： Wenkai Chang Guodong Yang Zhengxing Wu Zize Liang University of Chinese Academy of Sciences Beijing China State Key Laboratory of Management and Control for Complex Systems Institute of Automation Chinese Academy of Sciences Beijing China

Neural networks always require extensive training samples. However, in some special applications, i.e., insulators in high power grid, it is very hard and costly to collect variety-rich samples. In this study, a synthetic method is proposed to generate segmentation training samples for the insulators. Instead of relying on full-fledged Computer Graphic, this study focuses on the training features of neural networks. Based on this synthetic approach, many kinds of insulators samples including positive, empty and fake ones can be constructed, and their quantities are particularly balanced by an equalization strategy. In order to validate these produced samples, three end-to-end segmentation networks are employed to adapt to the generators in an adversarial training framework. Meanwhile, an improved training strategy is utilized to speed up the convergence. Finally, extensive experiments are executed to further analyze the proposed synthetic method and demonstrate its effectiveness for insulators segmentation.

关键词： Insulators Training Image segmentation Feature extraction Inspection Generators Neural networks

来源：评论

学校读者我要写书评

暂无评论

Overview of Image Segmentation and Its Application on Free Space Detection

Overview of Image Segmentation and Its Application on Free S...

引用

IEEE Data Driven control and Learning systems Conference

作者： Xiaodong Zhao Qichao Zhang Dongbin Zhao Zhonghua Pang The State Key Laboratory of Management and Control for Complex Systems Institute of Automation Chinese Academy of Sciences University of Chinese Academy of Sciences North China University of Technology

ISBN: (纸本)9781538626191

With the development of deep learning technique, image segmentation has received spreading attention in the computer vision field. It has a wide range of applications such as scene understanding, autonomous driving and so on. For the image segmentation, we can divide it into the semantic segmentation and instance segmentation, where a high-quality segmentation label for each instance is required for the latter method. In this paper, we sort out the popular structures of semantic segmentation and introduce the instance segmentation briefly. In the experiments, three main semantic segmentation methods are tested and analyzed based on the opened CamVid dataset, and the experiments for free space detection based on two popular segmentation methods are given.

关键词： Image segmentation Semantic segmentation Instance segmentation Free space detection

来源：评论

学校读者我要写书评

暂无评论

Automatic Guidewire Tip Segmentation in 2D X-ray Fluoroscopy Using Convolution Neural Networks

Automatic Guidewire Tip Segmentation in 2D X-ray Fluoroscopy...

引用

International Joint Conference on Neural Networks

作者： Yu-Dong Wu Xiao-Liang Xie Gui-Bin Bian Zeng-Guang Hou Xiao-Ran Cheng Sheng Chen Shi-Qi Liu Qiao-Li Wang University of Chinese Academy of Sciences Beijing China State Key Laboratory of Management and Control for Complex Systems Institute of Automation Chinese Academy of Sciences Beijing China

Guidewire tip detection in the percutaneous coronary intervention is important. It assists physicians in navigating and is a prerequisite for clinic applications such as surgical skill assessment and robot assisted surgery. Nevertheless, accurate detection is not a trivial task due to the noisy background of the 2D X-ray image and the thin, deformable structure of the tip. In this paper, an automatic method based on cascaded convolution neural networks is proposed to segment the tip in the 2D X-ray image. The main contribution of the method is to use a cascade detection-segmentation structure to overcome the noisy background and the large deformation of the tip, achieve robust, high-precision segmentation. On the other hand, sufficient annotated training samples are necessary for convolution neural network models, while pixel-level annotating is tedious and time-consuming. Accordingly, a novel data augmentation algorithm is introduced to improve the model generalization and performance, reduce the cost of data annotation. Evaluations were conducted on a dataset consisting of 22 different sequences of 2D X-ray images, 15 sequences for training and 7 sequences for evaluation. The proposed approach obtained tip precision of 0.532 pixels, F_1 score of 0.939, false tracking rate of 0.800%, and missing tracking rate of 9.900% on the test set. And the running speed is 4-5 frames per second.

关键词： Image segmentation Two dimensional displays X-ray imaging Catheters Convolution Training Strain

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：