This paper proposes a new method of semi-supervised human action recognition. In our approach, the motion energy image(MEI) and motion history image(MHI) are firstly used as the feature representation of the human action. Then, the constrained semi-supervised kmeans clustering algorithm is utilized to predict the class label of unlabeled training example. Meanwhile the average motion energy and history images are calculated as the recognition model for each category action. The category of the observed action is determined according to the correlation coefficients between its feature images and the pre-established average templates. The experiments on Weizmann dataset demonstrate that our method is effective and the average recognition accuracy can reach above 90% even when only using very small number of labeled action sequences.

A novel semi-supervised algorithm based on co-training is proposed in this paper. In the method, the motion energy history image are used as the different feature representation of human action; then the co-training based semi-supervised learning algorithm is utilized to predict the category of unlabeled training examples. And the average motion energy and history images are calculated as the recognition model for each category action. When recognition, the observed action is firstly classified through its correlation coefficients to the prior established templates respectively; then its final category is determined according to the consistency between the classification results of motion energy and motion history images. The experiments on Weizmann dataset demonstrate that our method is effective for human action recognition.

To accommodate the great challenge of obtaining accurate and detailed annotations of training data in human action recognition,in this paper we propose a semi-supervised algorithm based on Markov random walk.We first calculate mesh features of each image in the human action sequence.Then the mesh features are vector quantized through a rival penalized competitive neural network and the actions described by time-sequential images are converted into symbolic sequences.We then construct a Markov chain to reveal the labeled actions,unlabeled actions and action categories in training set according to their edit distance.Furthermore,we predict class label of unlabeled actions through a procedure of Markov random walk.The test sequence is then classified by the maximum posteriori probability criteria.The experiments on Weizmann dataset demonstrate the effectiveness of our method.The average recognition accuracy can exceed 80% even when only very small amount of labeled actions acquired.

A novel human action recognition algorithm based on edit distance is proposed in this paper. In the method, the mesh feature of each image in human action sequence is firstly calculated; then the feature vectors are quantized through a rival penalized competitive neural network; and through this processing, the time-sequential image sequences are converted into symbolic sequences. For human action recognition, the observed action is firstly vector quantized with the former competitive neural network; then the normalized edit distances to the training samples are calculated and the action which best matches the observed sequence is chosen as the final category. The experiments on Weizmann dataset demonstrate that our method is effective for human action recognition. The average recognition accuracy can reach above 94%.

In order to improve the performance of VoD,a hybrid server/P2P streaming system called xBT is introduced for large-scale VoD(Video-on-Demand)service,which can reduce delay time for play and mitigate the workload of servers.Drawbacks of current BitTorrent protocol for VoD application in a P2P mesh-based network is analyzed and improved the piece selection strategy used in BitTorrent to make it work more efficiently when VCR operations happened.The new piece selection strategy is based on the cold-prediction algorithm.The experiments indicate that xBT can enable the improvement of users' experiences in VoD by reducing delay and lower the overhead for servers in large scale VoD system.

The basic concepts and the relating theorem were studied in this paper.Knowledge reduction is one of the most important contents of rough set theory.The attribute reduction algorithm based on the frequency of attributes used the degree of the attributes appear in the discernibility matrix as the heuristic information of attributes selection.According to the limitation of this algorithm,an improved algorithm was presented.Finally the validity of the improved algorithm was analyzed and demonstrated by the example.

Along with the constant development of online services,the application of traditional RBAC model for the user-role assignment and maintaining the user-role assignment become an arduous and error-prone task.In order to solve these problems,this paper proposed one trust based user-role assignment model for assign role to users.It was based on the user trustworthiness in the system,which is a fuzzy concept.This paper provided a new method for the user-role assignment.

元数据是数据库,数据仓库中的核心部分,它是描述数据的数据.在一个信息系统中,可能存在不同的软件产品,每个软件产品可能使用了不同标准的元数据.为了使一系列给定的软件产品能够有效地参与ISC,并在数据层上进行互操作,就必须在元数据层上能有效地集成.所以,有必要研究元数据集成体系结构,分析元数据集成体系结构的关键点,并分析比较了几种解决方案.

In this paper, we analysis and research the intelligence home appliance based on pheromones theory. For the first time the theory of the pheromones of the ant colony algorithm used in intelligent learning of the Intelligence home appliance System. After a period of use, the system can accurately determine the user's operation, and suggest what user should to do automatically?and accurately. Besides, the system uses two types of communication as GSM and Bluetooth, so users can remotely control the family of Electrical and Electronic Equipment. This concentration of pheromone-based intelligent learning system has greatly enhanced the effectiveness and accuracy of the intelligence control.

Publish/Subscribe middleware system is the characteristics of synchronism, loose coupling and multi-point communications. At large-scale distribution system it has been growing wide applications. Efficient and practical matching algorithm is one of the key problems in the P/S system. Set up data structures about a event collection and a subscription collection, multi-tier index technology and the coverage relationship among events and subscriptions reduce the duplicated matching.

Contraposing the problem of the Multiple Access Interference (MAI) and Intersymbol Interference (ISI) caused by data transmission in channel, this study researches on multicarrier spread spectrum technique. According to the special structure of Multicarrier Direct Sequence Code Division Multiple Access system, this study put forward a mode of sending and receiving signals and gave the optimization of system parameters. By the simulating and comparing of the same bandwidth and measure arithmetic at last, it can obtain better performance than the Direct Sequence CDMA and greatly promote the capacity of anti-MAI and anti-ISI in the process of transmission.

Based on the Web services provided by data warehouse in power system,this paper analyzes the application of data warehouse in power utilities.It discusses the technologies needed and puts forward a solution of SOA-based data warehouse web services.It is of benefits to the information integration in power utilities.

Wind speed forecasting plays a significant role in the operation of wind power plants and power systems. An accurate forecasting on wind speed can effectively relieve or avoid the negative impact of wind power plants on power systems and enhance the competition of wind power plants in electric power market. Based on data mining, a method of wind speed forecasting is presented in this paper. By mining historical data as the learning stylebook using the fuzzy neural network (FNN), we can forecast the wind speed. The simulation results show that this method can improve the accuracy of wind speed forecasting effectively.

C ontraposing to the resources information sharing and cooperative work problems in power transformer cooperative design, we proposed a novel transformer cooperative design system architecture based on C/GS (Client/Grid Service) model on grid, by analyzing the communication model of B/S and C/S. Using Globus Toolkit realized Collaborative Design grid service, and proposed a conflict elimination method based on Rule Reasoning to deal with various conflict problems. This technology provides senior grid service and friendly graphical inter-face for the distributed places of designers, which has a tremendous degree to realize resource sharing and collaborative work.

Electric load forecasting is an important and challengeable work. In order to accurately forecast the loads of power system,this article presents a new short-term load forecasting method based on optimized decision tree,which efficiently takes the non-load factors’ influences into account. After preprocessing the sample data,rough set is used to reduce the testing attributes of decision tree. Good test results using actual data demonstrate that this method could improve the accuracy of short-term load forecasting ef-fectively,and it has practicability and superiority.

Concerning about the method of traditional color histogram is liable to lose some space information and sensitive to color diluting, so we make some modification to gray histogram to remedy the shortages mentioned above. As histogram is a well-known basic measurement to image color or gray layout, the paper makes similarity measures of different histograms based on their statistics peculiarity. The result of experience states that the method in my paper gets more improvement on the effect in the image retrieval.

To resolve the problem of the frequency selective fading and intersymbol interference (ISI) caused by data transmission in channel, this study researches on multicarrier spread spectrum technique. According to the special structure of multicarrier direct sequence code division multiple access (MC-DS-CDMA,) system, this study analyses the relation of parameters configure and resumable path numbers, and the performance of system in different parameters configure and multi-user. By the simulating and comparing of the same bandwidth and measure arithmetic at last, it can obtain better performance than the direct sequence CDMA (DS-CDMA, direct sequence code division multiple access) and greatly promote the capacity of anti-MAI (multiple access interference) and anti-ISI in the process of transmission.

Real-time data warehouse can provide enterprises with both strategic and tactical decision support. In this paper, we discussed several real-time data warehouse implementation scheme and proposed viable real-time data warehouse architecture, which capture change data by monitoring online log file. In this architecture, we use the dynamic multi-level cache to store and transmit data. This architecture realizes the data drip loading and optimizes the real-time data query and analyze.

The attribute reduction and relative attribute reduction were discussed in this paper. They are the core of KDD. The information view and the algebra view of rough set theory were combined and a novel attribute reduction algorithm was proposed. In the algorithm, the core attribute set which is the initial candidate reduction set is obtained from the discernibility matrix. The frequency of attributes, got from the filtered discernibility matrix, is used as the heuristic information of attributes selection. The algorithmpsilas terminal condition is realized by the conditional entropy. Taking the climatic factor reduction in load forecasting as an example, it has proved that the algorithm requires less computation, has high efficiency and can reduce the redundant attribute in the relative reduction set to a certain extent.

In this article, based on several real-time data warehouse implementation scheme, a SOA based real-time data warehouse architecture was proposed, which uses the SOA technology. In this architecture, the changed data was captured by the web service. And the dynamic multi-level cache was used for the real-time data drip loading and conflicts solving. The use of SOA and multi-level cache achieves data drip-loading for real-time data warehouse, facilitates the system's expanding and provides a new method for the construction of real-time data warehouse.