Model selection is important in many areas of supervised learning. Given a dataset and a set of models for predicting with that dataset, we must choose the model which is expected to best predict future data. In some situations, such as online learning for control of robots or factories, data is cheap and human expertise costly. Cross validation can then be a highly effective method for automatic model selection. Large scale cross validation search can, however, be computationally expensive. This paper introduces new algorithms to reduce the computational burden of such searches. We show how experimental design methods can achieve this, using a technique similar to a Bayesian version of Kaelbling’s Interval Estimation. Several improvements are then given, including (1) the use of blocking to quickly spot near-identical models, and (2) schemata search: a new method for quickly finding families of relevant features. Experiments are presented for robot data and noisy synthetic datasets. The new algorithms speed up computation without sacrificing reliability, and in some cases are more reliable than conventional techniques. 1

Selecting a good model of a set of input points by cross validation is a computationally intensive process, especially if the number of possible models or the number of training points is high. Techniques such as gradient descent are helpful in searching through the space of models, but problems such as local minima, and more importantly, lack of a distance metric between various models reduce the applicability of these search methods. Hoeffding Races is a technique for finding a good model for the data by quickly discarding bad models, and concentrating the computational effort at differentiating between the better ones. This paper focuses on the special case of leave-one-out cross validation applied to memorybased learning algorithms, but we also argue that it is applicable to any class of model selection problems. 1 Introduction Model selection addresses "high level" decisions about how best to tune learning algorithm architectures for particular tasks. Such decisions include which...

Given a set of models and some training data, we would like to find the model that best describes the data. Finding the model with the lowest generalization error is a computationally expensive process, especially if the number of testing points is high or if the number of models is large. Optimization techniques such as hill climbing or genetic algorithms are helpful but can end up with a model that is arbitrarily worse than the best one or cannot be used because there is no distance metric on the space of discrete models. In this paper we develop a technique called "racing" that tests the set of models in parallel, quickly discards those models that are clearly inferior and concentrates the computational effort on differentiating among the better models. Racing is especially suitable for selecting among lazy learners since training requires negligible expense, and incremental testing using leave-one-out cross validation is efficient. We use racing to select among various lazy learnin...

In this paper we introduce new algorithms for optimizing noisy plants in which each experiment is very expensive. The algorithms build a global non-linear model of the expected output at the same time as using Bayesian linear regression analysis of locally weighted polynomial models. The local model answers queries about confidence, noise, gradient and Hessians, and use them to make automated decisions similar to those made by a practitioner of Response Surface Methodology. The global and local models are combined naturally as a locally weighted regression. We examine the question of whether the global model can really help optimization, and we extend it to the case of time-varying functions. We compare the new algorithms with a highly tuned higher-order stochastic optimization algorithm on randomly-generated functions and a simulated manufacturing task. We note significant improvements in total regret, time to converge, and final solution quality. 1 INTRODUCTION In a stochastic optim...

The central thesis of this article is that memory-based methods provide natural and powerful mechanisms for high-autonomy learning control. This paper takes the form of a survey of the ways in which memory-based methods can and have been applied to control tasks, with an emphasis on tasks in robotics and manufacturing. We explain the various forms that control tasks can take, and how this impacts on the choice of learning algorithm. We show a progression of five increasingly more complex algorithms which are applicable to increasingly more complex kinds of control tasks. We examine their empirical behavior on robotic and industrial tasks. The final section discusses the interesting impact that explicitly remembering all previous experiences has on the problem of learning control.

Learning how to adjust to an opponent's position is critical to the success of having intelligent agents collaborating towards the achievement of specific tasks in unfriendly environments. This paper describes our work on developing methods to learn to choose an action based on a continuous-valued state attribute indicating the position of an opponent. We use a framework in which teams of agents compete in a simulator of a game of robotic soccer. We introduce a memory-based supervised learning strategy which enables an agent to choose to pass or shoot in the presence of a defender. In our memory model, training examples affect neighboring generalized learned instances with different weights. We conduct experiments in which the agent incrementally learns to approximate a function with a continuous domain. Then we investigate the question of how the agent performs in nondeterministic variations of the training situations. Our experiments indicate that when the random variations fall within some bound of the initial training, the agent performs better with some initial training rather than from a tabula-rasa.

We describe a density-adaptive reinforcement learning and a density-adaptive forgetting algorithm. This learning algorithm uses hybrid k-D/2k-trees to allow foravariable resolution partitioning and labelling of the input space. The density adaptive forgetting algorithm deletes observations from the learning set depending on whether subsequent evidence is available in a local region of the parameter space. The algorithms are demonstrated in a simulation for learning feasible robotic grasp approach directions and orientations and then adapting to subsequent mechanical failures in the gripper. 1

Model learning combined with dynamic programming has been shown to be effective for learning control of continuous state dynamic systems. The simplest method assumes the learned model is correct and applies dynamic programming to it, but many approximators provide uncertainty estimates on the fit. How can they be exploited? This paper addresses the case where the system must be prevented from having catastrophic failures during learning. We propose a new algorithm adapted from the dual control literature and use Bayesian locally weighted regression models with stochastic dynamic programming. A common reinforcement learning assumption is that aggressive exploration should be encouraged. This paper addresses the converse case in which the system has to reign in exploration. The algorithm is illustrated on a 4 dimensional simulated control problem. 1 Introduction Reinforcement learning and related grid-based dynamic programming techniques are increasingly being applied to dynamic system...

Autonomous navigation in outdoor, off-road environments requires solving complex classification problems. Obstacle detection, road following and terrain classification are examples of tasks which have been successfully approached using supervised machine learning techniques for classification. Large amounts of training data are usually necessary in order to achieve satisfactory generalization. In such cases, manually labeling data becomes an expensive and tedious process.

Given a set of models and some training data, we would like to find the model which best describes the data. Finding the model with the lowest generalization error is a computationally expensive process, especially if the number of testing points is high or if the number of models is large. Optimization techniques such as hill climbing or genetic algorithms are helpful but can end up with a model which is arbitrarily worse than the best one (local minima) or even worse, cannot be used because of the lack of a distance metric on the space of discrete models. In this paper we develop a technique called "racing" which tests the set of models in parallel, quickly discarding those models which are clearly inferior and concentrating the computational effort on differentiating among the better models. We use racing to select among various memory based models and to find relevant features in applications ranging from robot juggling to lesion detection in MRI scans.