The Bayesian "evidence" approximation has recently been employed to determine the noise and weight-penalty terms used in back-propagation. This paper shows that for neural nets it is far easier to use the exact result than it is to use the evidence approximation. Moreover, unlike the evidence approximation,the exact result neither has to be re-calculated for every new data set, nor requires the running of computer code (the exact result is closed form). In addition, it turns out that the evidence procedure's MAPestimate for neural nets is, in toto, approximation error. Another advantage of the exact analysis is that it does not lead one to incorrect intuition, like the claim that using evidence one can "evaluate different priors in light of the data". This paper also discusses sufficiency conditions for the evidence approximation to hold, why it can sometimes give "reasonable" results, etc.

We present a methodological framework enabling a detailed description ofthe performance of Hopfield-like attractor neural networks (ANN) in the first two iterations. Using the Bayesian approach, wefind that performance is improved when a history-based term is included in the neuron's dynamics. A further enhancement of the network's performance is achieved by judiciously choosing the censored neurons (those which become active in a given iteration) onthe basis of the magnitude of their post-synaptic potentials. Thecontribution of biologically plausible, censored, historydependent dynamicsis especially marked in conditions of low firing activity and sparse connectivity, two important characteristics of the mammalian cortex. In such networks, the performance attained ishigher than the performance of two'independent' iterations, whichrepresents an upper bound on the performance of history-independent networks.

My Ph.D. dissertation describes PAGODA (probabilistic autonomous goal-directed agent), a model for an intelligent agent that learns autonomously in domains containing uncertainty. The ultimate goal of this line of research is to develop intelligent problem-solving and planning systems that operate in complex domains, largely function autonomously, use whatever knowledge is available to them, and learn from their experience. PAGODA was motivated by two specific requirements: The agent should be capable of operating with minimal intervention from humans, and it should be able to cope with uncertainty (which can be the result of inaccurate sensors, a nondeterministic environment, complexity, or sensory limitations). I argue that the principles of probability theory and decision theory can be used to build rational agents that satisfy these requirements.

This paper presents a simple framework for Horn-clause abduction, with probabilities associated with hypotheses. The framework incorporates assumptions about the rule base and independence assumptions amongst hypotheses. It is shown how any probabilistic knowledge representable in a discrete Bayesian belief network can be represented in this framework. The main contribution is in finding a relationship between logical and probabilistic notions of evidential reasoning. This provides a useful representation language in its own right, providing a compromise between heuristic and epistemic adequacy. It also shows how Bayesian networks can be extended beyond a propositional language.

English-to-French I f _[ French-to-English We present an overview of Candide, a system for automatic e Channel " -] Decoder 6 translation of French text to English text.

In this paper we investigate an average-case model of concept learning, and give results that place the popular statistical physics and VC dimension theories of learning curve behavior in a common framework.

"Best-first model merging" is a general technique for dynamically choosing the structure of a neural or related architecture while avoiding overfitting. It is applicable to both leaming and recognition tasks and often generalizes significantly better than fixed structures. We demonstrate the approach applied to the tasks of choosing radial basis functions for function learning, choosing local affine models for curve and constraint surface modelling, and choosing the structure of a balltree or bumptree to maximize efficiency of access.

The Bayesian model comparison framework is reviewed, and the Bayesian Occam's razor is explained. This framework can be applied to feedforward networks, making possible (1) objective comparisons between solutions using alternative network architectures; (2) objective choice of magnitude and type of weight decay terms; (3) quantified estimates of the error bars on network parameters and on network output. The framework also generates a measure of the effective number of parameters determined by the data. The relationship of Bayesian model comparison to recent work on prediction of generalisation ability (Guyon et al., 1992, Moody, 1992) is discussed.

In a Bayesian framework, we give a principled account of how domainspecific prior knowledge such as imperfect analytic domain theories can be optimally incorporated into networks of locally-tuned units: by choosing a specific architecture and by applying a specific training regimen. Our method proved successful in overcoming the data deficiency problem in a large-scale application to devise a neural control for a hot line rolling mill. It achieves in this application significantly higher accuracy than optimally-tuned standard algorithms such as sigmoidal backpropagation, and outperforms the state-of-the-art solution.

The Bayesian model comparison framework is reviewed, and the Bayesian Occam's razor is explained. This framework can be applied to feedforward networks, making possible (1) objective comparisons between solutions using alternative network architectures; (2) objective choice of magnitude and type of weight decay terms; (3) quantified estimates of the error bars on network parameters and on network output. The framework also generates a measure of the effective number of parameters determined by the data. The relationship of Bayesian model comparison to recent work on prediction of generalisation ability (Guyon et al., 1992, Moody, 1992) is discussed.