This paper introduces Evolutionary Pursuit (EP) as a novel and adaptive representation method for image encoding and classification. In analogy to projection pursuit methods, EP seeks to learn an optimal basis for the dual purpose of data compression and pattern classification. The challenge for EP is to increase the generalization ability of the learning machine as a result of seeking the trade-off between minimizing the empirical risk encountered during training and narrowing the confidence interval for reducing the guaranteed risk during future testing on unseen images. Towards that end, EP implements strategies characteristic of genetic algorithms (GAs) for searching the space of possible solutions to determine the optimal basis. EP starts by projecting the original data into a lower dimensional whitened Principal Component Analysis (PCA) space. Directed but random rotations of the basis vectors in this space are then searched by GAs where evolution is driven by a fitness function defined in terms of performance accuracy (`empirical risk') and class separation (`confidence interval'). Accuracy indicates the extent to which learning has been successful so far, while separation gives an indication of the expected fitness on future trials. The feasibility of the new method has been successfully tested on face recognition where the large number of possible bases requires some type of greedy search algorithm. The particular face recognition task involves 1,107 FERET frontal face images corre-

The agent-based approach emphasizes the importance of learning through organism-environment interaction. This approach is part of a recent trend in computational models of learning and development toward studying autonomous organisms that are embedded in virtual or real environments. In this paper we introduce the concepts of online and offline sampling and highlight the role of online sampling in agent-based models. After comparing the strengths of each approach for modeling particular developmental phenomena and research questions, we describe a recent agent-based model of infant causal perception. We conclude by discussing some of the present limitations of agent-based models and suggesting how these challenges may be addressed. © 2001 Academic Press Computational models of learning and development are playing an increasingly critical role in child development research (Cassidy, 1990;

Abstract not found

We present a novel evolutionary approach to robotic control of a real robot based on genetic programming (GP). Our approach uses genetic programming techniques that manipulate machine code to evolve control programs for robots. This variant of GP has several advantages over a conventional GP system, such as higher speed, lower memory requirements and better real time properties. Previous attempts to apply GP in robotics use simulations to evaluate control programs and have difficulties with learning tasks involving a real robot. We present an on-line control method that is evaluated in two different physical environments and applied to two tasks using the Khepera robot platform: obstacle avoidance and object following. The results show fast learning and good generalization.

This paper presents an approach to solve the symbol grounding problem within the framework of embodied cognitive science. It will be argued that symbolic structures can be used within the paradigm of embodied cognitive science by adopting an alternative definition of a symbol. In this alternative definition, the symbol may be viewed as a structural coupling between an agent's sensorimotor activations and its environment. A robotic experiment is presented in which mobile robots develop a symbolic structure from scratch by engaging in a series of language games. In this experiment it is shown that robots can develop a symbolic structure with which they can communicate the names of a few objects with a remarkable degree of success. It is further shown that, although the referents may be interpreted differently on different occasions, the objects are usually named with only one form.

Abstract-Many forms of learning depend on the ability of an organism to sense and react to the adaptive value of its behavior. Such value, if reflected in the activity of specific neural structures (neural value systems), can selectively increase the probability of adaptive behaviors by modulating synaptic changes in the circuits relevant to those behaviors. Neuromodulatory systems in the brain are well suited to carry out this process since they respond to evolutionarily important cues (innate value), broadcast their responses to widely distributed areas of the brain through diffuse projections, and release substances that can modulate changes in synaptic strength. The main aim of this paper is to show that, if value-dependent modulation is extended to the inputs of neural value systems themselves, initially neutral cues can acquire value. This process has important implications for the acquisition of behavioral sequences. We have used a synthetic neural model to illustrate value-dependent acquisition of a simple foveation response to a visual stimulus. We then examine the improvement that ensues when the connections to the value system are themselves plastic and thus become able to mediate acquired value. Using a second-order conditioning paradigm, we demonstrate that auditory discrimination can occur in the model in the absence of direct positive reinforcement and even in the presence of slight negative reinforcement. The discriminative responses are accompanied by

Abstract not found

This article describes a theory of the computations underlying the selection of coordinated motion patterns, especially in reaching tasks. The central idea is that when a spatial target is selected as an object to be reached, stored postures are evaluated for the contributions they can make to the task. Weights are assigned to the stored postures, and a single target posture is found by taking a weighted sum of the stored postures. Movement is achieved by reducing the distance between the starting angle and target angle of each joint. The model explains compensation for reduced joint mobility, tool use, practice effects, performance errors, and aspects of movement kinematics. Extensions of the model can account for anticipation and coarticulation effects, movement through via points, and hierarchical control of series of movements. The goal of this research is a unified theory of the planning and control of physical action. Such a theory, as several authors have noted (Jeannerod, in press; Rosenbaum, 1991; Wing, 1993), has been lacking. Instead, specialized models have been designed to account for data from different tasks. The sentiment

On behalf of:

This thesis discusses a biologically defensible model of development for artificial organisms. It can be used in conjunction with genetic algorithms to design autonomous agents, complete with body and nervous system. It also has potential applications in the field of theoretical biology. The approach taken is one of exploration, i.e. the model is implemented simultaneously at three different levels of complexity: regulation of gene expression, development of body morphology and finally neural development. At each level the model's behavior under evolutionary pressure is addressed. The whole integrated model will be illustrated with a hand-designed artificial organism, that is capable of executing a simple avoidance behavior in a simulated world.