Yu-Xiang Zhao

SOM-based optimization

A new approach to optimization problems based on the self-organizing feature maps is proposed. We name the new optimization algorithm the SOM-based optimization (SOMO) algorithm. Through the self-organizing process, good solutions to an optimization problem can be simultaneously explored and exploited. An additional advantage of the algorithm is that the outputs of the neural network allow us to transform a multi-dimensional fitness landscape into a three-dimensional projected fitness landscape. Several simulations are used to illustrate the effectiveness of the proposed optimization algorithm.

A swarm-inspired projection algorithm

In this paper a new data projection algorithm which was inspired by the foraging behaviors of doves is proposed. We name the new data projection the swarm-inspired projection (SIP) algorithm. The algorithm allows us to visually estimate the number of clusters existing in a data set. Based on the projection result, we may then partition the data set into the corresponding number of clusters. The SIP algorithm regards each data pattern in a data set as a piece of crumb which will be sequentially tossed to a flock of doves on the ground. The doves will adjust their physical positions to compete for crumbs. Gradually, the flock of doves will be divided into several groups according to the distributions of the crumbs. The formed groups will naturally correspond to the underlying data structures in the data set. By viewing the scatter plot of the final positions of the doves we can estimate the number of clusters existing in the data set. Several data sets were used to demonstrate the effectiveness of the proposed SIP algorithm.

A variant of the SOM algorithm and its interpretation in the viewpoint of social influence and learning

The conventional self-organizing feature map (SOM) algorithm is usually interpreted as a computational model, which can capture main features of computational maps in the brain. In this paper, we present a variant of the SOM algorithm called the SOM-based optimization (SOMO) algorithm. The development of the SOMO algorithm was motivated by exploring the possibility of applying the SOM algorithm in continuous optimization problems. Through the self-organizing process, good solutions to an optimization problem can be simultaneously explored and exploited by the SOMO algorithm. In our opinion, the SOMO algorithm not only can be regarded as a biologically inspired computational model but also may be regarded as a new approach to a model of social influence and social learning. Several simulations are used to illustrate the effectiveness of the proposed optimization algorithm.

Application of SOMO Based Clustering in Building Renovation

Building renovations are usually performed as required based on inconvenience or damage that has already taken place. Construction practitioners are seldom aware of the relationships between all the related factors and their corresponding costs. The purpose of this study is to apply the self-organizing feature map (SOM) optimization based clustering (SOMOC) algorithm to building renovations so as to evaluate its feasibility and provide solutions. We collected 1056 sets of building renovation data sampled from 102 buildings. The SOMOC algorithm is utilized to expose the tendency in view of basic building features. The results suggest that the SOMOC method is feasible and effectively divides the data into 8 clusters for cluster analysis. In the subsequent discussion, findings imply that: (1) all clusters have similar distributions in terms of proportion of building age and building size, and thus, no rule can be formed for renovation practice; and (2) location, structure type, renovation frequency and cost are all related to each other. The benefits of the study not only prove the practicability of SOMOC but help the construction practitioners to learn from the past.

A NEURAL-NETWORK-BASED APPROACH TO RECOGNIZING 3D ARM MOVEMENTS

Gesture recognition is needed for a variety of applications. One particular application of gesture-based systems is to implement a speaking aid for the deaf. Among several factors constituting a hand gesture, the arm movement pattern is one of the most challenging features to recognize. In this paper, we propose a neural-network-based approach to recognition of spatio-temporal patterns of nonlinear 3D arm movements. Compared to Hidden-Markov-Model-based methods, the most appealing property of the proposed method is its simplicity. The effectiveness of this method is evaluated by a database consisted of 10 persons.

A Simple Approach to Stereo Matching and Its Application in Developing a Travel Aid for the Blind

In this paper, we present an idea of using stereo matching to develop a travel aid for the blind. In this approach, images are segmented into several nonoverlapping homogeneous regions using a color segmentation algorithm. For each homogeneous region, a rectangular window, which is large enough to cover the region, is found. A local match with the found rectangular window size is then executed to find the disparity for the considered region. A clustering algorithm is adopted to cluster the disparities into several major different values. Finally, a piece-wise disparity map is constructed. Based on the disparity map, information about the unfamiliar environments in front of the blind will be output to them. With the information about the environment the blind will have less fear in walking through unfamiliar environments via white canes.

Feature Selection Method Based on Neighborhood Relationships: Applications in EEG Signal Identification and Chinese Character Recognition.

In this study, a new feature selection algorithm, the neighborhood-relationship feature selection (NRFS) algorithm, is proposed for identifying rat electroencephalogram signals and recognizing Chinese characters. In these two applications, dependent relationships exist among the feature vectors and their neighboring feature vectors. Therefore, the proposed NRFS algorithm was designed for solving this problem. By applying the NRFS algorithm, unselected feature vectors have a high priority of being added into the feature subset if the neighboring feature vectors have been selected. In addition, selected feature vectors have a high priority of being eliminated if the neighboring feature vectors are not selected. In the experiments conducted in this study, the NRFS algorithm was compared with two feature algorithms. The experimental results indicated that the NRFS algorithm can extract the crucial frequency bands for identifying rat vigilance states and identifying crucial character regions for recognizing Chinese characters.

To develop the Mandarin-phonetic-symbol communication aid for high-functioning autism children

In this study, a Mandarin-phonetic-symbol communication aid named as zhuyin communication board is developed for children with high-functioning autism. The zhuyin communication board can execute on Tablet PC to assists autistic children to express their thought to other people. When an autistic child wants to express his thought, he can press the corresponding phonetic symbols by using the developed zhuyin communication board. To motivate the interesting of autistic children, the developed app provides a picture-based testing for learning Mandarin phonetic symbols of variety objects. Compared with the traditional paper keyboard, the developed aid could show the typing phonetic symbol immediately on the screen, and provides the voice of zhuyin pronunciation to improve autistic childrens language perception skill.

Implementation of Parallel Computing FAST Algorithm on Mobile GPU

Corner detection is an extremely important technique in image recognition, which is widely employed in various applications for image recognition. With the widespread use of mobile devices, image recognition techniques are frequently applied in such devices. However, the hardware resource of smartphones is lacking and restricted; it is a difficult task to apply the techniques of corner detection smoothly in these devices. To enhance the computational speed, the FAST corner detection algorithm is implemented with parallel computing of GPU in mobile devices. In the experiments, the computational speed of the FAST corner detection algorithm increases 24 times after using GPU parallel computing. Compared with the widely known SURF algorithm, which is computed with mobile CPU only, the proposed technique in this study is 468 times faster than SURF algorithm.

To Develop the Virtual Physics Laboratory by Integrating Kinect with Gesture Classification Algorithm

Physics is an experimental science, which is through experiments for initiating students into physical concepts and principles. To motivate students in learning physics, in this study, a virtual physics laboratory was developed by using the techniques of Kinect, Unity3D and a gesture classification algorithm. The visual physics experiments were designed in the virtual physics laboratory. The experimental results show that the user can accurately interact with the virtual objects in the virtual physics laboratory, and the developed system provides an interesting way to assist students in learning physics.