The S-Map is a network with a simple learning algorithm that combines theself-organization capability of the Self-Organizing Map (SOM) and the probabilistic interpretability of the Generative Topographic Mapping(GTM). The simulations suggest that the S Map algorithm has a stronger tendency to self-organize from random initialconfiguration than the GTM. The S-Map algorithm can be further simplified to employ pure Hebbian learning, without changingthe qualitative behaviour of the network. 1 Introduction The self-organizing map (SOM; for a review, see [1]) forms a topographic mapping from the data space onto a (usually two-dimensional) output space. Most of these questions arise because of the following two facts: the SOM is not a generative model, i.e. it does not generate a density in the data space, and it does not have a well-defined objective function that the training process would strictly minimize. However, it seems that the GTM requires a careful initialization to self-organize.

We implement a model of obstacle avoidance in flying insects on a small, monocular robot. The result is a system that is capable of rapid navigation through a dense obstacle field. The key to the system is the use of zigzag behavior to articulate the body during movement. It is shown that this behavior compensates for a parallax blind spot surrounding the focus of expansion normally foundin systems without parallax behavior.

"Boosting" is a general method for improving the performance of any learning algorithm that consistently generates classifiers which need to perform only slightly better than random guessing. A recently proposed and very promising boosting algorithm is AdaBoost [5]. It has been applied withgreat success to several benchmark machine learning problems using rather simple learning algorithms [4], and decision trees [1, 2, 6]. In this paper we use AdaBoost to improve the performances of neural networks. We compare training methods based on sampling the training set and weighting the cost function. Our system achieves about 1.4% error on a data base of online handwritten digits from more than 200 writers. Adaptive boosting of a multi-layer network achieved 1.5% error on the UCI Letters and 8.1 % error on the UCI satellite data set.

A new policy iteration algorithm for partially observable Markov decision processes is presented that is simpler and more efficient than an earlier policy iteration algorithm of Sondik (1971,1978). The key simplification is representation of a policy as a finite-state controller. This representation makes policy evaluation straightforward. The paper's contributionis to show that the dynamic-programming update used in the policy improvement step can be interpreted as the transformation ofa finite-state controller into an improved finite-state controller. The new algorithm consistently outperforms value iteration as an approach to solving infinite-horizon problems.

Some learning techniques for classification tasks work indirectly, by first trying to fit a full probabilistic model to the observed data. Whether this is a good idea or not depends on the robustness with respect to deviations from the postulated model. We study this question experimentally in a restricted, yet nontrivial and interesting case: we consider a conditionally independent attribute (CIA) model which postulates a single binary-valued hidden variable z on which all other attributes (i.e., the target and the observables) depend. In this model, finding the most likely value of anyone variable (given known values for the others) reduces to testing a linear function of the observed values. We learn CIA with two techniques: the standard EM algorithm, and a new algorithm we develop based on covariances. We compare these, in a controlled fashion, against an algorithm (a version of Winnow) that attempts to find a good linear classifier directly. Our conclusions help delimit the fragility of using the CIA model for classification: once the data departs from this model, performance quickly degrades and drops below that of the directly-learned linear classifier.

This paper enhances the Q-Iearning algorithm for optimal asset allocation proposedin (Neuneier, 1996 [6]). The new formulation simplifies the approach by using only one value-function for many assets and allows model-freepolicy-iteration. After testing the new algorithm on real data, the possibility of risk management within the framework of Markov decision problems is analyzed. The proposed methods allows the construction of a multi-period portfolio management system which takes into account transaction costs, the risk preferences of the investor, and several constraints on the allocation. 1 Introduction

An emerging use of reinforcement learning (RL) is to approximate optimal policies for large-scale control problems through extensive simulated control experience. Described here are initial experiments directed toward the development of an automated recovery system (ARS)for high-agility aircraft. An ARS is an outer-loop flight control system designed to bring the aircraft from a range of initial states to straight, level, and non-inverted flight in minimum time while satisfying constraints such as maintaining altitude and accelerations within acceptable limits. Here we describe the problem and present initial results involving only single-axis (pitch) recoveries. Through extensive simulated control experience using a medium-fidelity simulation of an F-16, the RL system approximated an optimal policy for longitudinal-stick inputs to produce near-minimum-time transitions to straight and level flight in unconstrained cases, as well as while meeting a pilot-station acceleration constraint. 2 AIRCRAFT MODEL

In most neural network models, synapses are treated as static weights that change only on the slow time scales of learning. In fact, however, synapses are highly dynamic, and show use-dependent plasticity over a wide range of time scales. Moreover, synaptic transmission is an inherently stochastic process: a spike arriving at a presynaptic terminal triggers release of a vesicle of neurotransmitter from a release site with a probability that can be much less than one. Changes in release probability represent one of the main mechanisms by which synaptic efficacy is modulated in neural circuits. We propose and investigate a simple model for dynamic stochastic synapses that can easily be integrated into common models for neural computation. We show through computer simulations and rigorous theoretical analysis that this model for a dynamic stochastic synapse increases computational power in a nontrivial way. Our results may have implications for the processing oftime-varying signals by both biological and artificial neural networks. A synapse 8 carries out computations on spike trains, more precisely on trains of spikes from the presynaptic neuron. Each spike from the presynaptic neuron mayor may not trigger the release of a neurotransmitter-filled vesicle at the synapse.

A simple linear averaging of the outputs of several networks as e.g. in bagging [3], seems to follow naturally from a bias/variance decomposition of the sum-squared error. The sum-squared error of the average model is a quadratic function of the weighting factors assigned to the networks in the ensemble [7], suggesting a quadratic programming algorithm for finding the "optimal" weighting factors. If we interpret the output of a network as a probability statement, the sum-squared error corresponds to minus the loglikelihood or the Kullback-Leibler divergence, and linear averaging of the outputs tologarithmic averaging of the probability statements: the logarithmic opinion pool. The crux of this paper is that this whole story about model averaging, bias/variancedecompositions, and quadratic programming to find the optimal weighting factors, is not specific for the sumsquared error,but applies to the combination of probability statements of any kind in a logarithmic opinion pool, as long as the Kullback-Leibler divergence plays the role of the error measure. As examples we treat model averaging for classification models under a cross-entropy error measure and models for estimating variances.

Motivated by the findings of modular structure in the association cortex, we study a multi-modular model of associative memory that can successfully store memory patterns with different levels of activity. Weshow that the segregation of synaptic conductances into intra-modular linear and inter-modular nonlinear ones considerably enhances the network's memory retrieval performance. Compared with the conventional, single-module associative memory network, the multi-modular network has two main advantages: It is less susceptible todamage to columnar input, and its response is consistent with the cognitive data pertaining to category specific impairment. 1 Introduction Cortical modules were observed in the somatosensory and visual cortices a few decades ago. These modules differ in their structure and functioning but are likely to be an elementary unit of processing in the mammalian cortex. Within each module the neurons are interconnected.