Hamid Gholam Hosseini

Facial expression analysis for estimating learner's emotional state in intelligent tutoring systems

Intelligent tutoring systems (ITS) provide individualized instruction. They offer many advantages over the traditional classroom scenario: they are always available, nonjudgmental and provide tailored feedback resulting in increased and effective learning. However, they are still not as effective as one-on-one human tutoring. The next generation of intelligent tutors is expected to be able to take into account the cognitive and emotional state of students. We present a proposed contribution of affect to student modeling, and reports on the progress made in the development of a facial expression analysis component for intelligent tutoring systems.

Intelligent Fish Freshness Assessment

Fish species identification and automated fish freshness assessment play important roles in fishery industry applications. This paper describes a method based on support vector machines (SVMs) to improve the performance of fish identification systems. The result is used for the assessment of fish freshness using artificial neural network (ANN). Identification of the fish species involves processing of the images of fish. The most efficient features were extracted and combined with the down-sampled version of the images to create a 1D input vector. Max-Win algorithm applied to the SVM-based classifiers has enhanced the reliability of sorting to 96.46%. The realisation of Cyranose 320 Electronic nose (E-nose), in order to evaluate the fish freshness in real-time, is experimented. Intelligent processing of the sensor patterns involves the use of a dedicated ANN for each species under study. The best estimation of freshness was provided by the most sensitive sensors. Data was collected from four selected species of fishes over a period of ten days. It was concluded that the performance can be increased using individual trained ANN for each specie. The proposed system has been successful in identifying the number of days after catching the fish with an accuracy of up to 91%.

Fault signal propagation in a network of distributed motors

Industrial environment usually contains multiple motors that are supplied through a common power bus. The power line acts as a good conducting environment for signals to travel through the power network. In effect, this influences other motors with noisy signals that may indicate a fault condition. Further complexity arises when signals are generated by motors with different power ratings, a different slip speed and more than one source of fault signals. This sort of complexity and mixing among signals from multiple sources makes them difficult to measure and precisely correlate to a given machine or fault. In this research, a power network model for induction motors is presented. This model accommodates the signal attenuation factor when it is disseminated along the bus. It also allows for system configuration of motor characteristics and fault injection. It is expected that this simulation model will facilitate the environment for testing sensor networks and data fusion approaches that facilitates better intelligence for fault identification and localization.

Machine learning-based clinical decision support system for early diagnosis from real-time physiological data

This research aims to design a self-organizing decision support system for early diagnosis of key physiological events. The proposed system consists of pre-processing, clustering and diagnostic system, based on self-organizing fuzzy logic modeling. The clustering technique was employed with empirical pattern analysis, particularly when the information available is incomplete or the data model is affected by vagueness, which is mostly the case with medical/clinical data. Clustering module can be viewed as unsupervised learning from a given dataset. This module partitions the patient vital signs to identify the key relationships, patterns and clusters among the medical data. Secondly, it uses self-organizing fuzzy logic modeling for early symptom and event detection. Based on the clustering outcome, when detecting abnormal signs, a high level of agreement was observed between system interpretation and human expert diagnosis of the physiological events and signs.

A Secure Wireless Telehealthcare Monitoring System and its Web Application

Efforts to prepare for a growing number of elderly patients, reducing the escalation of healthcare costs, and avoiding hospitals emergency room overcrowding are some of the driving forces for adopting wireless healthcare monitoring systems. However, due to the open-to-air commination nature of multilayer wireless networks, it is important to consider reliability, accuracy, security and privacy of such data transmission. We have developed a low-cost and wireless telehealthcare system for monitoring of basic physiological parameters and automatically transmitting the measured data to an electronic patient record. It employs off the shelf wireless products and a secure web-based application which have been tested in a hospital with satisfactory outcomes.

Detecting and Identifying Industrial Gases by a Method Based on Olfactory Machine at Different Concentrations

Gas sensors have been widely reported for industrial gas detection and monitoring. However, the rapid detection and identification of industrial gases are still a challenge. In this work, we measure four typical industrial gases including CO2, CH4, NH3, and volatile organic compounds (VOCs) based on electronic nose (EN) at different concentrations. To solve the problem of effective classification and identification of different industrial gases, we propose an algorithm based on the selective local linear embedding (SLLE) to reduce the dimensionality and extract the features of high-dimensional data. Combining the Euclidean distance (ED) formula with the proposed algorithm, we can achieve better classification and identification of four kinds of gases. We compared the classification and recognition results of classical principal component analysis (PCA), linear discriminate analysis (LDA), and PCA + LDA algorithms with the proposed SLLE algorithm after selecting the original data and performing feature extraction. The experimental results show that the recognition accuracy rate of the SLLE reaches 91.36%, which is better than the other three algorithms. In addition, the SLLE algorithm provides more efficient and accurate responses to high-dimensional industrial gas data. It can be used in real-time industrial gas detection and monitoring combined with gas sensor networks.

A novel voice signal discrimination algorithm and its application

A new algorithm which is capable of classifying voice signals recorded from subjects before and after the anesthetic procedure is presented. This new algorithm is based on Wavelet Packet Analysis (WPA) and Least Squares Support Vector Machines (LSSVM) and it combines the coefficients of WPA with other parameters, such as Spectral Centroid, Spectral roll-off point etc., as the feature vector. Experimental evaluation has shown that the proposed classification algorithm based on WPA and LSSVM is very effective as compared to the other two methods, and the total accuracy rate is over 85%.