This paper describes the new features available in the Sim-Point 3.0 release. The release provides two techniques for drastically reducing the run-time of SimPoint: faster searching to find the best clustering, and efficiently clustering large numbers of intervals. SimPoint 3.0 also provides an option to output only the simulation points that represent the majority of execution, which can reduce simulation time without much increase in error. Finally, this release provides support for correctly clustering variable length intervals, taking into consideration the weight of each interval during clustering. This paper describes SimPoint 3.0’s new features, how to use them, and points out some common pitfalls. 1

We describe an implemented system that learns to recognize human faces under varying pose and illumination conditions. The system relies on symmetry operations to detect the eyes and the mouth in a face image, uses the locations of these features to normalize the appearance of the face, performs simple but effective dimensionality reduction by a convolution with a set of Gaussian receptive fields, and subjects the vector of activities of the receptive fields to a Radial Basis Function interpolating classifier. The performance of the system compares favorably with the state of the art in machine recognition of faces.

Partially observable Markov decision processes (POMDPs) can be used to model complex control problems that include both action outcome uncertainty and imperfect observability. A control problem within the POMDP framework is expressed as a dynamic optimization problem with a value function that combines costs or rewards from multiple steps. Although the POMDP framework is more expressive than other simpler frameworks, like Markov decision processes (MDP), its associated optimization methods are more demanding computationally and only very small problems can be solved exactly in practice. Our work focuses on two possible approaches that can be used to solve larger problems: approximation methods and exploitation of additional problem structure. First, a number of new eÆcient approximation methods and improvements of existing algorithms are proposed. These include (1) the fast informed bound method based on approximate dynamic programming updates that lead to piecewise linear and convex v...

this paper we investigate the problem of providing error bounds for approximation of an unknown function from scattered, noisy data. This problem has particular relevance in the field of machine learning, where the unknown function represents the task that has to be learned and the scattered data represents the examples of this task. An obvious quantity of interest for us is the generalization error -- a measure of how much the result of the approximation scheme differs from the unknown function -- typically studied as a function of the number of data points. Since the data are randomly generated and noisy, the analysis of the generalization error necessarily involves statistical considerations in addition to the traditional

Training a multilayer neural net by back-propagation is slow and requires arbitrary choices regarding the number of hidden units and layers. This paper describes an algorithm which is much faster than back-propagation and for which it is not necessary to specify the number of hidden units in advance. The relationship with other fast pattern recognition algorithms, such as algorithms based on k-d trees, is mentioned. The algorithm has been implemented and tested on articial problems such as the parity problem and on real problems arising in speech recognition. Experimental results, including training times and recognition accuracy, are given. Generally, the algorithm achieves accuracy as good as or better than nets trained using back-propagation, and the training process is much faster than back-propagation. Accuracy is comparable to that for the \nearest neighbour" algorithm, which is slower and requires more storage space. Comments Only the Abstract is given here. The full paper ap...

This paper examines four current theoretical approaches to the representation and recognition of visual objects: structural descriptions, geometric constraints, multidimensional feature spaces, and shape-space approximation. The strengths and the weaknesses of the theories are considered, with a special focus on their approach to categorization --- a computationally challenging task which is not widely addressed in computer vision (where the stress is rather on the generalization of recognition across changes of viewpoint).

This paper describes using multi-agent reinforcement learning (RL) algorithms for learning traffic light controllers to minimize the overall waiting time of cars in a city. The RL systems learn value functions estimating expected waiting times for cars given different settings of traffic lights. Selected settings of traffic lights result from combining the predicted waiting times of all cars involved. We investigate RL systems using different kinds of global communicated information between traffic light agents. We also show how the value functions can be used by the driving policies of cars to select optimal routes to destination addresses. The experimental results show that the RL algorithms can outperform non-adaptable traffic light controllers, and that optimizing driving policies is very useful.

Psychophysical findings accumulated over the past several decades indicate that perceptual tasks such as similarity judgment tend to be performed on a low-dimensional representation of the sensory data. Low dimensionality is especially important for learning, as the number of examples required for attaining a given level of performance grows exponentially with the dimensionality of the underlying representation space. In this chapter, we argue that, whereas many perceptual problems are tractable precisely because their intrinsic dimensionality is low, the raw dimensionality of the sensory data is normally high, and must be reduced by a nontrivial computational process, which, in itself, may involve learning. Following a survey of computational techniques for dimensionality reduction, we show that it is possible to learn a low-dimensional representation that captures the intrinsic low-dimensional nature of certain classes of visual objects, thereby facilitating further learning of tasks...

The nearest neighbor technique is a simple and appealing method to address classification problems. It relies on the assumption of locally constant class conditional probabilities. This assumption becomes invalid in high dimensions with a finite number of examples due to the curse of dimensionality. We propose a technique that computes a locally flexible metric by means of Support Vector Machines (SVMs). The maximum margin boundary found by the SVM is used to determine the most discriminant direction over the query's neighborhood. Such direction provides a local weighting scheme for input features. We present experimental evidence of classification performance improvement over the SVM algorithm alone and over a variety of adaptive learning schemes, by using both simulated and real data sets.

We present an efficient “sparse sampling ” technique for approximating Bayes optimal decision making in reinforcement learning, addressing the well known exploration versus exploitation tradeoff. Our approach combines sparse sampling with Bayesian exploration to achieve improved decision making while controlling computational cost. The idea is to grow a sparse lookahead tree, intelligently, by exploiting information in a Bayesian posterior—rather than enumerate action branches (standard sparse sampling) or compensate myopically (value of perfect information). The outcome is a flexible, practical technique for improving action selection in simple reinforcement learning scenarios. 1.