The use of multilayer perceptrons (MLP) with threshold functions (binary step function activations) greatly reduces the complexity of the hardware implementation of neural networks...
Vassilis P. Plagianakos, George D. Magoulas, Micha...
Hidden Markov models assume that observations in time series data stem from some hidden process that can be compactly represented as a Markov chain. We generalize this model by as...
The European Network for Lifelong Competence Development is an Integrated Project funded by the EU. Its goal is to establish an innovative technical and organizational infrastructu...
Abstract. We develop a probabilistic interpretation of non-linear component extraction in neural networks that activate their hidden units according to a softmaxlike mechanism. On ...
Visualization and interactivity are valuable active learning techniques that can improve mastery of difficult concepts. In this paper we describe jFAST, an easy-to-use graphical s...