AI Magazine, 13 (1), 32-44.

A quantitative and practical Bayesian framework is described for learning of mappings in feedforward networks. The framework makes possible (1) objective comparisons between solutions using alternative network architectures, (2) objective stopping rules for network pruning or growing procedures, (3) objective choice of magnitude and type of weight decay terms or additive regularizers (for penalizing large weights, etc.), (4) a measure of the effective number of well-determined parameters in a model, (5) quantified estimates of the error bars on network parameters and on network output, and (6) objective comparisons with alternative learning and interpolation models such as splines and radial basis functions. The Bayesian "evidence" automatically embodies "Occam's razor," penalizing overflexible and overcomplex models. The Bayesian approach helps detect poor underlying assumptions in learning models. For learning models well matched to a problem, a good correlation between generalization ability and the Bayesian evidence is obtained.

We describe a representation and set of inference techniques for the dynamic construction of probabilistic and decision-theoretic models expressed as networks. In contrast to probabilistic reasoning schemes that rely on fixed models, we develop a representation that implicitly encodes a large number of possible model structures. Based on a particular query and state of information, the system constructs a customized belief net for that particular situation. We develop an interpretation of the network construction process in terms of the implicit networks encoded in the database. A companion method for constructing belief networks with decisions and values (decision networks) is also developed that uses sensitivity analysis to focus the model building process. Finally, we discuss some issues of control of model construction and describe examples of constructing networks.

Architecture) provides a generic domain independent computational architecture suitable for command, planning and execution applications.

Assessing probabilities in such situations seems difficult.

J. International Computer Chess Association, 15 (1): 16-18.

We argue that the problem of plan recognition, inferring an agent's plan from observations, is largely a problem of inference under conditions of uncertainty. We present an approach to the plan recognition problem that is based on Bayesian probability theory. In attempting to solve a plan recognition problem we first retrieve candidate explanations. These explanations (sometimes only the most promising ones) are assembled into a plan recognition Bayesian network, which is a representation of a probability distribution over the set of possible explanations. We perform Bayesian updating to choose the most likely interpretation for the set of observed actions.

Creating an optimum long-term schedule for the Hubble Space Telescope is difficult by almost any standard due to the large number of activities, many relative and absolute time constraints, prevailing uncertainties and an unusually wide range of timescales. This problem has motivated research in neural networks for scheduling. The novel concept of continuous suitaility functions defined over a continuous time domain has been developed to represent soft temporal relationships between activities. All constraints and preferences are automatically translated into the weights of an appropriately designed artificial neural network. The constraints are subject to propagation and consistency enhancement in order to increase the number of explicitly represented constraints.

"Work-in-progress on the design of a conditional nonlinear planner is described. CNLP is a nonlinear planner that develops plans that account for foreseen uncertainties. CNLP represents an extension of the conditional planning technique of Warren [75] to the domain of nonlinear planning." In ICAPS-92, pp. 189–197.

A computational scheme for reasoning about dynamic systems using (causal) probabilistic networks is presented. The scheme is based on the framework of Lauritzen and Spiegel-halter (1988), and may be viewed as a generalization of the inference methods of classical time-series analysis in the sense that it allows description of non-linear, multivariate dynamic systems with complex conditional independence structures. Further, the scheme provides a method for efficient backward smoothing and possibilities for efficient, approximate forecasting methods. The scheme has been implemented on top of the HUGIN shell.