Yoon-Seok Yang

Data Refinement and Channel Selection for a Portable E-Nose System by the Use of Feature Feedback

We propose a data refinement and channel selection method for vapor classification in a portable e-nose system. For the robust e-nose system in a real environment, we propose to reduce the noise in the data measured by sensor arrays and distinguish the important part in the data by the use of feature feedback. Experimental results on different volatile organic compounds data show that the proposed data refinement method gives good clustering for different classes and improves the classification performance. Also, we design a new sensor array that consists only of the useful channels. For this purpose, each channel is evaluated by measuring its discriminative power based on the feature mask used in the data refinement. Through the experimental results, we show that the new sensor array improves both the classification rates and the efficiency in computation and data storage.

LDA-Based Vapor Recognition Using Image-Formed Array Sensor Response for Portable Electronic Nose

The efficient formulation of measured data in artificial olfactory system is very important for simplicity, robustness and implementation of the algorithm especially in portable system which has limited resources. In this study, we applied the linear discriminant analysis (LDA) to E-nose measurements formulated in 2 dimensional matrices. The 160 measurements for 8 different vapors using 6 channel sensor array were identified with the accuracy of 98.75%. LDA is one of the most efficient computer vision algorithms to recognize image objects. It maintained the significant features for vapor classification during dimension reduction. This can simplify further processing like storage or transmission. Therefore, the proposed method will help the realization of the ubiquitous or embedded olfactory sensing system.

Multi-dimensional vibration energy harvester for efficient use in common environment

This study proposes a novel type of energy harvester for more effective harvesting of multi-dimensional vibration existing in common environment such as car, bicycle, human movements, etc. The proposed harvester has multi-dimensionally triggered vibrating components in a toroidal electromagnetic coupling structure. We designed and implemented a prototype harvester and compared its output characteristic with conventional cylindrical electromagnetic harvester under three different angle conditions which correspond to three orthogonal axes of multi-dimensional vibration. The results show that the proposed doughnut-shaped energy harvester operate under more broad range of vibration angle than conventional cylindrical counterpart where. Further study will focus on the improvement of output performance and availability of the proposed harvester in common environment beyond laboratory conditions.

Time Horizon Selection Using Feature Feedback for the Implementation of an E-Nose System

Several aspects of sensors should be considered for their practical application in reducing the cost and improving the performance of the system. Some factors, such as the power consumption, sampling period, processing time and memory size, are particularly important for efficient portable systems. In this paper, we propose a time horizon selection method for a portable sensor system. By using the feature feedback to investigate the relation between the input space and feature space, we find the distribution of discriminant information in a data sample and distinguish the time horizon with which we can improve the performance of the classification. The experimental results on different volatile organic compounds show that the proposed method provides for the good clustering of the different classes and increases classification rates from 95.3% to 96.9% and 98.2% for the selected time horizons, respectively.

The Efficiency of Feature Feedback Using R-LDA with Application to Portable E-Nose System

In this paper, we improve the performance of Feature Feedback and present its application for vapor classification in a portable E-Nose system. Feature Feedback is a preprocessing method which detects and removes unimportant information from input data so that classification performance is improved. In our original Feature Feedback algorithm, PCA is used before LDA in order to avoid the small sample size (SSS) problem but it is said that this may cause loss of significant discriminant information for classification. To overcome this, in the proposed method, we improve Feature Feedback using regularized Fisher’s separability criterion to extract the features and apply it to E-Nose system. The experimental result shows that the proposed method works well.

Development of the brain stimulator for stroke recovery using ZigBee technology

We studied the prototypal developments of Plastic Cortex Stimulator (PCS) for stroke recovery. The PCS implanted in damaged cortex of rat brain can increase the plasticity of nearby cortical area, so that the motor sensation can be improved. For future implantation, an wireless technology is necessary.

Pegboard Evaluation Automation Utilizing RFID System for Telerehabilitation

This study proposes an automated pegboard utilizing the RFID system with multiple reader antennas for the rehabilitation services and the occupational therapy. The system automates the scoring by detecting the plugging correctness as well as the plugging status. It also aims to increase the patient`s interest and the functional intelligence. The system was prototyped and the RFID read rate (over 99.998%) was confirmed. The system was also tested for the automatic capability of the scoring the session. The proposed system will be served as the typical example for the ubiquitous rehabilitation devices.