The procedure of calculating Mel Frequency based Cepstral Coefficients (MFCC) is shown to resemble a three layer Multilayer Perceptron (MLP) like structure. Such an MLP is employed as a preprocessor in a hybrid HMM-MLP system, and the possibility of optimizing the whole system as a single entity, with respect to a suitable criterion, is pointed out. This system, together with the Maximum Mutual...

In this paper, we address the problem of how many randomly labeled patterns can be correctly classified by a single-layer perceptron when the patterns are correlated with each other. In order to solve this problem, two analytical schemes are developed based on the replica method and Thouless-Anderson-Palmer (TAP) approach by utilizing an integral formula concerning random rectangular matrices. ...

We present a novel method for finding low dimensional views of high dimensional data: Targeted Projection Pursuit. The method proceeds by finding projections of the data that best approximate a target view. Two versions of the method are introduced; one version based on Procrustes analysis and one based on a single layer perceptron. These versions are capable of finding orthogonal or nonorthogo...

Breast tissue microarrays (TMAs) facilitate the study of very large numbers of breast tumours in a single histological section, but their scoring by pathologists is time consuming, typically highly quantised, and not without error. This paper compares the results of different classification and ordinal regression algorithms trained to predict the scores of immunostained breast TMA spots, based ...

Rosenblatt's convergence theorem for the simple perceptron initiated much excitement about iterative weight modifying neural networks. However, this convergence only holds for the class of linearly separable functions, which is vanishingly small compared to arbitrary functions. With multilayer networks of nonlinear units it is possible, though not guaranteed, to solve arbitrary functions. Backp...

Calculation and analysis of annual R-factor local values for 103 stations in Poland were the main aims of this study. Calculations were made by means of single hidden layer perceptron artificial neural network on the base of monthly precipitation totals from years: 1961-1980. For most of the analyzed stations calculated average annual R-factor values were low or moderate, at the range from 50 t...

We study the interaction between input distributions, learning algorithms and nite sample sizes in the case of learning classiication tasks. Focusing on the case of normal input distributions, we use statistical mechanics techniques to calculate the empirical and expected (or generalization) errors for several well-known algorithms learning the weights of a single-layer perceptron. In the case ...

In previous work, it has been experimentally shown that the implementation of Error Correcting Output Coding (ECOC) classification methods with an ensemble of parallel and independent non linear dichotomizers (ECOC PND) outperforms the implementation with a single monolithic multi layer perceptron (ECOC MLP). This result was ascribed to the higher effectiveness of error correcting output coding...

We present context-sensitive Multiple Task Learning, or csMTL as a method of inductive transfer embedded in the well known WEKA machine learning suite. csMTL uses a single output neural network and additional contextual inputs for learning multiple tasks. Inductive transfer occurs from secondary tasks to the model for the primary task so as to improve its predictive performance. The WEKA multi-...

Weight decay was proposed to reduce over tting as it often appears in the learning tasks of arti cial neural networks. In this paper weight decay is applied to a well de ned model system based on a single layer perceptron, which exhibits strong over tting. Since the optimal non-over tting solution is known for this system, we can compare the effect of the weight decay with this solution. A stra...