Presents results of the face verification contest that was organized in conjunction with International Conference on Pattern Recognition 2000. Participants had to use identical data sets from a large, publicly available multimodal database XM2VTSDB. Training and evaluation was carried out according to an a priori known protocol. Verification results of all tested algorithms have been collected and made public on the XM2VTSDB website, facilitating large scale experiments on classifier combination and fusion. Tested methods included, among others, representatives of the most common approaches to face verification -elastic graph matching, Fisher's linear discriminant and support vector machines.

Describes a new, logic-based methodology for analyzing observations. The key features of this “logical analysis of data” (LAD) methodology are the discovery of minimal sets of features that are necessary for explaining all observations and the detection of hidden patterns in the data that are capable of distinguishing observations describing “positive” outcome events from “negative” outcome events. Combinations of such patterns are used for developing general classification procedures. An implementation of this methodology is described in this paper, along with the results of numerical experiments demonstrating the classification performance of LAD in comparison with the reported results of other procedures. In the final section, we describe three pilot studies on applications of LAD to oil exploration, psychometric testing and the analysis of developments in the Chinese transitional economy. These pilot studies demonstrate not only the classification power of LAD but also its flexibility and capability to provide solutions to various case-dependent problems.

Biometric person identity authentication is gaining more and more attention. The authentication task performed by an expert is a binary classification problem: reject or accept identity claim. Combining experts, each based on a different modality (speech, face, fingerprint, etc.), increases the performance and robustness of identity authentication systems. In this context, a key issue is the fusion of the different experts for taking a final decision (i.e., accept or reject identity claim). We propose to evaluate different binary classification schemes (support vector machine, multilayer perceptron, C4.5 decision tree, Fisher's linear discriminant, Bayesian classifier) to carry on the fusion. The experimental results show that support vector machines and Bayesian classifier achieve almost the same performances, and both outperform the other evaluated classifiers.

This paper describes how a speaker verification task can be advantageously decomposed into a series of binary classification problems, i.e. each problem discriminating between two classes only. Each binary classifier is specific to one speaker, one anti-speaker and one word. Decision trees dealing with attributes of continuous values are used as classifiers. The set of classifiers is then pruned to eliminate the less relevant ones. Diverse pruning methods are experimented, and it is shown that when the speaker verification decision is performed with an a priori threshold, some of them give better results than a reference HMM system.

Quantization of the parameters of a Perceptron is a central problem in hardware implementation of neural networks using a numerical technology. A neural model with each weight limited to a small integer range will require little surface of silicon. Moreover, according to Occam's razor principle, better generalization abilities can be expected from a simpler computational model. The price to pay for these benefits lies in the difficulty to train these kind of networks. This paper proposes essentially two new ideas for constructive training algorithms, and demonstrates their efficiency for the generation of feedforward networks composed of Boolean threshold gates with discrete weights. A proof of the convergence of these algorithms is given. Some numerical experiments have been carried out and the results are presented in terms of the size of the generated networks and of their generalization abilities.

Linear threshold Boolean units (LTUs) are the basic processing components of artificial neural networks of Boolean activations. Quantization of their parameters is a central question in hardware implementation, when numerical technologies are used to store the configuration of the circuit. In the previous studies on the circuit complexity of feedforward neural networks, no differences had been made between a network with “small” integer weights and one composed of majority units (LTUs with weights in $\{ - 1,0, + 1\} $), since any connection of weight w (w integer) can be simulated by $|w|$ connections of value sgn(w). This paper will focus on the circuit complexity of democratic networks, i.e., circuits of majority units with at most one connection between each pair of units. The main results presented are the following: any Boolean function can be computed by a depth-3 nondegenerate democratic network and can be expressed as a linear threshold function of majorities; AT-LEAST-k and AT-MOST-k are computable by a depth-2, polynomial-sized democratic network; the smallest sizes of depth-2 circuits computing PARITY are identical for a democratic network and for a usual network; the VC-dimension of the class of the majority functions is $n + 1$, i.e., equal to that of the class of any linear threshold functions.