Sameer Singh

Fuzzy neural computing of coffee and tainted-water data from an electronic nose

Abstract In this paper we compare the ability of a fuzzy neural network and a common back-propagation network to classify odour samples that were obtained by an electronic nose employing semiconducting oxide conductometric gas sensors. Two different sample sets have been analysed: first, the aroma of three blends of commercial coffee, and secondly, the headspace of six different tainted-water samples. The two experimental data sets provide an excellent opportunity to test the ability of a fuzzy neural network due to the high level of sensor variability often experienced with this type of sensor. Results are presented on the application of three-layer fuzzy neural networks to electronic nose data. They demonstrate a considerable improvement in performance compared to a common back-propagation network.

Advances in Pattern Recognition — ICAPR 2001

Two novel concepts in structural pattern recognition are discussed in this paper. The rst, median of a set of graphs, can be used to characterize a set of graphs by just a single prototype. Such a characterization is needed in various tasks, for example, in clustering. The second novel concept is weighted mean of a pair of graphs. It can be used to synthesize a graph that has a speci ed degree of similarity, or distance, to each of a pair of given graphs. Such an operation is needed in many machine learning tasks. It is argued that with these new concepts various well-established techniques from statistical pattern recognition become applicable in the structural domain, particularly to graph representations. Concrete examples include k-means clustering, vector quantization, and Kohonen maps.

2D spiral pattern recognition with possibilistic measures

The main task for a 2D spiral recognition algorithm is to learn to discriminate between data distributed on two distinct strands in the x-y plane. This problem is of critical importance since it incorporates temporal characteristics often found in real-time applications, i.e. the spiral coils with time. Previous work with this benchmark has witnessed poor results with statistical methods such as discriminant analysis and tedious procedures for better results with neural networks. This paper describes a fuzzy approach which outperforms previous work in terms of the recognition rate and the speed of recognition. The paper presents the new approach and results with the validation and test sets. The results show that it is possible to solve the spiral problem in a relatively small amount of time with the fuzzy approach (up to 100% correct classification on the validation and test set; 77.2% correct classification with cross-validation using the leave-one-out method).

A pattern matching tool for time-series forecasting

We describe a pattern recognition based tool for forecasting. We compare the results of forecasting with this tool against the exponential smoothing method on Santa Fe series data and US financial index. The results show that the pattern recognition based tool is highly accurate on standard error measures.

EFFECT OF NOISE ON GENERALISATION IN MASSIVELY PARALLEL FUZZY SYSTEMS

This paper studies the performance of Massively Parallel Fuzzy Systems (MPFS) on the two spiral benchmark. Spiral data is contaminated with five different noise distributions. The recognition rates of the system are reported with varying levels of different types of noise. The behaviour of the system is investigated with additive, multiplicative, cumulative and non-cumulative noise. The results show that the MPFS system remains stable to different noise variations and the generalisation error remains consistently low. As the total noise in the system increases, the system witnesses a linear decrease in entropy and the generalisation error is easier to predict. The error rate is found to have two separate patterns of variation for cumulative and non-cumulative noise.

Optical character recognition: Neural network analysis of hand-printed characters

The main objective of this paper is to introduce a novel method of feature extraction for character data and develop a neural network system for recognising different Latin characters. In this paper we describe feature extraction, neural network development for character recognition and perform further neural network analysis on noisy image segments to explain the qualitative aspects of handwriting.

Machine Recognition of Hand-Printed Chinese Characters

The recognition of Chinese characters has been an area of great interest for many years, and a large number of research papers and reports have already been published in this area. There are several major problems with Chinese character recognition: Chinese characters are distinct and ideographic, the character size is very large and many structurally similar characters exist in the character set. Thus, classification criteria are difficult to generate.This article presents a new technique for the recognition of hand-printed Chinese characters using statistical pattern classification. Conventional methods have relied on hand-constructed dictionaries which are tedious to construct, and difficult to make tolerant to variation in writing styles. The article also discusses Chinese character recognition using dominant point feature extraction, and statistical pattern classification. The system was tested with 500 characters each character has 40 samples, and the rate of recognition obtained was 84.45%. This strongly supports the usefulness of the proposed measures for Chinese character classification.

Classifier Systems Based on Possibility Distributions: A Comparative Study

The main aim of this paper is three fold: a) to understand the working of a classifier system based on possibility distribution functions, b) to evaluate its performance against other superior methods such as fuzzy and non-fuzzy neural networks on real data, c) and finally to recommend changes for enhancing its performance. The paper explains how to construct a possibility based classifier system which is used with conventional error-estimation techniques such as cross-validation and bootstrapping. The results were obtained on a set of electronic nose data and this performance was compared with earlier published results on the same data using fuzzy and non-fuzzy neural networks. The results show that the possibility approach is superior to the non-fuzzy approach, however, further work needs to be done.

Off-Line Handwritten Chinese Character Recognition Based on Structural Features and Fuzzy Artmap

Recognition of printed and handwritten text (characters and digits) has been a challenge for many researchers in the field of Document Image Analysis for almost thirty years. Many existing pattern recognition methods have been used and new ones have been proposed. Despite the satisfactory results obtained in recognizing printed text, recognition of handwritten text is still an open research. This paper presents a method for off-line recognition of handwritten Chinese characters. Each character is described by six structural features extracted using the dominant point method. The goodness of each feature is determined by a method called stroke probability. The classification was performed by the Fuzzy ARTMAP neural network within the context of the one-class problem approach. Evaluation of the proposed method was determined using database of 900 characters (an average of 40 samples/character). The average mean recognition rate was approximately 94%.

Quantitative classification of conversational language using artificial neural networks

Abstract In this paper I shall describe the use of artificial neural networks for the classification of subjects based on their conversational speech using a set of linguistic measures, with particular reference to the application of this approach in classifying dysphasic patients. These linguistic measures can be applied to the transcribed texts of conversational speech of both normal and dysphasic subjects, and will quantify the availability of linguistic features which are dependent on word frequency. The paper presents the results of a crossvalidation study using neural networks and compares them against those obtained using a linear discriminant analysis on the same data.