Most models of object recognition assume a convergent projection within a hierarchy of processing stages: Stimuli from different areas of the receptive field project onto the same population of cells. Since attention is likely to modify such spatial pooling by changing the competitive weight of...

An alternative model to class conditional probability estimation for classification problems is introduced. The model assumes a deterministic mapping between input space and the set of classes, training data consists of probability distributions of the classes at the measurements points. A...

This paper proposes a new architecture for neural networks where the hidden layer is completely replaced by what is referred to as a 'principal component layer'. This layer consists of only a few selected principal components that perform the function of hidden nodes. Then, an algorithm based on...

This paper presents a Hopfield neural network model for matching features invariant to projective transformation. The projective invariance has been firstly embedded into the compatibility constraint, such that the problem of finding point correspondences can be formulated as minimizing the...

It is well known that multilayer neural networks with generalized delta rule (GDR) can map any arbitrary functions. The relationship between statistical linear regression and network learning has been attributed to the principal components which can be extracted from the input patterns along....

The computational power of recurrent RBF (radial basis functions) networks is investigated. A recurrent network which consists of RBF units with maximum norm employing any activation function that has different values in at least two nonnegative points, is constructed so as to implement a given...

Generalised linear regression models are powerful regression models, often designed using weight decay regularisation to achieve good generalisation. he parameters associated with this method of regularisation have an interpretation as hyperparameters in a Bayesian framework. Standard approaches...

Automated incident detection is an essential component of a modern freeway traffic monitoring system. A number of neural network-based incident detection models have been tested independently over the past decade. This paper evaluates the adaptability of three promising neural network models for...

A denoising unit based on wavelet multiresolution analysis is added ahead of the multilayered perceptron. The cost function used in neural network learning is also applied as a denoising criterion and hence denoising itself is treated as a part of the uniform model. By introducing continuously...

This paper presents an algorithm of playing game tic-tac-toe. This algorithm is learned by a modified neural network (NN), which is trained by an improved genetic algorithm (GA). In the proposed NN, the neuron has two activation transfer functions and exhibits a node-to-node relationship in the...