Structure-Adaptable Digital Neural Networks

N euroengineers have came up with the idea of arti cial neural networks, massively parallel computing machines inspired by biological nervous systems. Such systems o er a new paradigm, where learning from examples or learning from interaction replaces programming. They are basically composed of interconnected simple processing elements (i.e., arti cial neurons or simply neurons). A predominant approach in the eld of arti cial neural networks consists of using a database to train the system by applying a so-called learning algorithm to modify the interconnection dynamics between arti cial neurons, using a predetermined network topology (i.e., the number of arti cial neurons and the way they are interconnected). When the training process ends, the system remains xed, and can be considered a learned system. However, in applications where the entire training database is not available or may change in time, the learning process must persist, giving rise to learning systems. A problem of such systems is how to preserve what has been previously learned while continuing to incorporate new knowledge. Learning systems overcome such problems by a local adaptation process and by o ering the possibility of dynamically modifying the network's topology. Most arti cial neural network applications today are executed as conventional software simulations on sequential machines with no dedicated hardware. Recon gurable hardware can provide the best features of both dedicated hardware circuits and software implementations: high-density and high-performance designs, and rapid prototyping and exibility. The main goal in this thesis was therefore the development of structure-adaptable digital neural networks with continual learning capabilities, using eld-programmable logic devices. Neural network models with modi able structure have already been developed, but are usually computationally intensive. Therefore, we have developed the FAST neural architecture, an unsupervised learning with Flexible Adaptable-Size Topology that does not require intensive computation to learn and recon gure its structure, and can thus exist as a stand-alone learning machine that can be used, for example, in real-time control applications, such as robot navigation. The FAST learning system was conceived to handle the problem of dynamic categorization or online clustering. In this thesis, we are interested in two di erent categorization tasks: in the rst task, related to image processing, the process of color categorization is used for image segmentation, while in the second task, related to neurocontrol, an autonomous intelligent system self-categorizes (or clusters) its sensor readings in order to integrate the relentless bombardment of signals coming from the