Shunxiang Wu

Feature selection based on quality of information

Feature selection as one of the key problems of data preprocessing is a hot research topic in pattern recognition, machine learning, and data mining. Evaluating the relevance between features based on information theory is a popular and effective method. However, very little research pays attention to the distinguishing ability of feature, i.e., the degree of a feature distinguishes a given sample with other samples. In this paper, we propose a new feature selection method based on the distinguishing ability of feature. First, we define the concept of maximum-nearest-neighbor, and use this concept to discriminate the nearest neighbors of samples. Then, we present a new measure method for evaluating the quality of feature. Finally, the proposed algorithm is tested on benchmark datasets. Experimental results show that the proposed algorithm can effectively select a discriminative feature subset, and performs as well as or better than other popular feature selection algorithms.

Research of an E-mail forensic and analysis system based on visualization

Nowadays, E-mail communication has been abused for numerous illegitimate purposes such as E-mail spamming, terrorist attack, business fraud, etc. As a result, to analysis the rich personal information hidden in E-mail is significant for investigation and evidence collection. In this paper, an investigation and analysis system aiming to Email was presented, which supports a variety of data sources including the preserved Email client data files, databases as well as text files. The system firstly parses related data files, preprocess the data, and then, a key word search technique based on KMP algorithm was adopted to classify the E-mail collections into different categories. Afterwards, an association frequency mining based on statistics will be performed to discover the association features behind email accounts. To make the forensic results more readable, we will associate the E-mail accounts with personnel information table in reality. The final forensic results will be visualized using related layout techniques to make the information more illustrative and understandable.

Cross-Platform App Recommendation by Jointly Modeling Ratings and Texts

Over the last decade, the renaissance of Web technologies has transformed the online world into an application (App) driven society. While the abundant Apps have provided great convenience, their sheer number also leads to severe information overload, making it difficult for users to identify desired Apps. To alleviate the information overloading issue, recommender systems have been proposed and deployed for the App domain. However, existing work on App recommendation has largely focused on one single platform (e.g., smartphones), while it ignores the rich data of other relevant platforms (e.g., tablets and computers). In this article, we tackle the problem of cross-platform App recommendation, aiming at leveraging users’ and Apps’ data on multiple platforms to enhance the recommendation accuracy. The key advantage of our proposal is that by leveraging multiplatform data, the perpetual issues in personalized recommender systems—data sparsity and cold-start—can be largely alleviated. To this end, we propose a hybrid solution, STAR (short for “croSs-plaTform App Recommendation”) that integrates both numerical ratings and textual content from multiple platforms. In STAR, we innovatively represent an App as an aggregation of common features across platforms (e.g., App’s functionalities) and specific features that are dependent on the resided platform. In light of this, STAR can discriminate a user’s preference on an App by separating the user’s interest into two parts (either in the App’s inherent factors or platform-aware features). To evaluate our proposal, we construct two real-world datasets that are crawled from the App stores of iPhone, iPad, and iMac. Through extensive experiments, we show that our STAR method consistently outperforms highly competitive recommendation methods, justifying the rationality of our cross-platform App recommendation proposal and the effectiveness of our solution.

Obstacle avoidance and formation regrouping strategy and control for UAV formation flight

Obstacle avoidance (OA) and formation regrouping (FR) are very important issues for implementation of automatic UAVs formation flight. In this paper, the OA strategy and control for UAV leader is investigated first. Then, it is followed by investigation of the OA strategy and control for UAV follower. In the process of OA, in order to conduct OA reasonably and efficiently, the formation is split first and then all the UAVs regroup to fly in the original formation after OA process is completed. To this end, formation splitting and regrouping strategies in four OA scenarios are presented for UAV formation flight. Simulation experimental results demonstrate that the presented OA and FR strategies and control are effective.

Rating LDA model for collaborative filtering

People are pleased with the great wealth of products in online stores. However, it is more and more difficult for people to choose their favorite products in an online store. Thus, recommendation systems are necessary to provide useful suggestions and selections. A users choice is not only influenced by his/her interests, but also by the ratings of others. In this paper, we propose a Rating LDA (RLDA) Model for collaborative filtering by adding rating information to the Latent Dirichlet Allocation (LDA). User behavior is not independent; it follows the trend of others. Therefore, we assume that for similar interests, the higher the proportion of high ratings, the more popular the items. We perform experiments on two real world data sets: MovieLens100k and MovieLens1M. Results show that, in terms of F1 score, our proposed approach significantly outperforms some baseline methods.

An improved K-means clustering method for cDNA microarray image segmentation.

Microarray technology is a powerful tool for human genetic research and other biomedical applications. Numerous improvements to the standard K-means algorithm have been carried out to complete the image segmentation step. However, most of the previous studies classify the image into two clusters. In this paper, we propose a novel K-means algorithm, which first classifies the image into three clusters, and then one of the three clusters is divided as the background region and the other two clusters, as the foreground region. The proposed method was evaluated on six different data sets. The analyses of accuracy, efficiency, expression values, special gene spots, and noise images demonstrate the effectiveness of our method in improving the segmentation quality.

Research and Design of the Differential Autonomous Mobile Robot Based on Multi-Sensor Information Fusion Technology

The composite array of the ultrasonic and infrared sensors is researched and designed to expand the scope of the robot detection. A novel approach of multi-sensor information fusion based on the neural networks and the fuzzy control is presented. A BP neural network is used to fuse the information from multi-sensor so that the uncertainty of the sensors’ information can be decreased and high accuracy of obstacle identification can be obtained. In order to realize the decision control, a fuzzy control technology is used for obstacle avoidance. The simulation from Mobotsim software shows that the information fusion method presented in this paper has high performance of robustness and flexibility when dealing with the obstacle avoidance problem and proved the effectiveness of the proposed approach. Keywords-Mobile robot; Multi-sensor information fusion; Obstacle avoidance; BP neural network; Fuzzy control

Research of Brushless DC Motor Simulation System Based on RBF-PID Algorithm

This paper has made a thorough analysis of Brushless DC Motor Simulation System, in which we use a single neuron PID control algorithm based on RBF neural network for on-line identification (RBF-PID Algorithm). A special human-computer interaction (HCI) interface was designed in this paper, which provides interface for users to set the correlative parameters and select related control algorithm. Two control algorithms: Conventional PID and RBF-PID Algorithm were respectively adopted in this paper to make a comparison. The result shows that RBF-PID Algorithm performs better controlling Brushless DC Motor. Thus, the Simulation System is of great advantage for parameters testing and setting for the controller of Brushless DC Motor compared with traditional manual methods in practical use.

Study of Grey Rough Set Model Based on Tolerance Relation

This paper analyses several extended rough set models in incomplete information systems and proposes a tolerance relation based model of processing grey incomplete information systems, which is an extension to rough set models. The method of the model is: firstly partitioning the original incomplete information system by introduced threshold value, then establishing tolerance classes through grey tolerance relation and obtaining upper and lower approximations through these tolerance classes. Moreover, a method of whitening grey numbers based on grey tolerance relation is given. This paper shows that the model accords with practice according to examples and the algorithm of whitening grey numbers is also comparatively ideal. The more important point is that the subjective needs are considered during partitioning grey tolerance classes by introducing threshold value. So it is consistent with the system methodology of person-oriented person-to-machine communication

UAV formation flight control and formation switch strategy

During the formation flight time, the Unmanned Aerial vehicles (UAVs) have to transform their flight formation in many cases, such as environmental change, the modification of the task and some UAVs leaving of the formation. During the time to change the formation, the most important thing is to avoid UAVs crash into each other, then, they should finish the transformation process in certain time. This paper firstly focuses on the UAV formation keeping by using Proportion Integration Differentiation (PID) control method. Secondly, several control strategy are designed for formation transformation. Then, use the above methods to complete the formation transformation. Finally, the effectiveness of UAVs formation transformation is taken into account by simulated verification and comparative analysis.