Learning programs that generalize from real‐world examples will have to deal with many different kinds of data. Continuous numeric data can cause problems for algorithms that search for examples with identical property values. These problems can be surmounted by categorizing the numeric data. However, this process has problems of its own. In this paper, we look at the need for categorizing numeric data and several methods for doing so. We concentrate on the use of generalization‐based memory, a memory organization where (...) actual examples are stored along with generalizations, which leads to a generalization‐based categorization algorithm. We also consider how to use a number heuristic, looking for gaps. These methods have been implemented in the UNIMEM computer system. Examples are presented of these algorithms categorizing data about the states of the United States. (shrink)

This paper presents a skill learning model CLARION. Different from existing models of mostly high-level skill learning that use a top-down approach (that is, turning declarative knowledge into procedural knowledge through practice), we adopt a bottom-up approach toward low-level skill learning, where procedural knowledge develops first and declarative knowledge develops later. Our model is formed by integrating connectionist, reinforcement, and symbolic learning methods to perform on-line reactive learning. It adopts a two-level dual-representation framework (Sun, 1995), with a combination of localist (...) and distributed representation. We compare the model with human data in a minefield navigation task, demonstrating some match between the model and human data in several respects. (shrink)

Machine learning has been proven useful for solving the bottlenecks in building expert systems. Noise in the training instances will, however, confuse a learning mechanism. Two main steps are adopted here to solve this problem. The first step is to appropriately arrange the training order of the instances. It is well known from Psychology that different orders of presentation of the same set of training instances to a human may cause different learning results. This idea is used here for machine (...) learning and an order arrangement scheme is proposed. The second step is to modify a conventional noise-free learning algorithm, thus making it suitable for noisy environment. The generalized version space learning algorithm is then adopted to process the training instances for deriving good concepts. Finally, experiments on the Iris Flower problem show that the new scheme can produce a good training order, allowing the generalized version space algorithm to have a satisfactory learning result. (shrink)

_role, especially in learning, and through devising hybrid neural network models that (in a qualitative manner) approxi-_ _mate characteristics of human consciousness. In doing so, the paper examines explicit and implicit learning in a variety_ _of psychological experiments and delineates the conscious/unconscious distinction in terms of the two types of learning_ _and their respective products. The distinctions are captured in a two-level action-based model C_larion_. Some funda-_ _mental theoretical issues are also clari?ed with the help of the model. Comparisons with (...) existing models of conscious-_. (shrink)

This article uses simulation as an empirical method for identifying process models of strategy effects in a category-learning task. A general set of learning assumptions defined a symbolic learning framework in which alternative simulation models were defined and tested. The goal was to identify process models that could account for previously reported data on the interaction between how a learner encounters category variance across a series of training samples and whether the task instructions suggested an active, hypothesis-testing approach, or a (...) more passive learning mode. Descriptive characterizations of active and passive learning were mapped into complementary settings of parameters operating with the general learning framework. Alternative models, defined by different configurations of these parameters, were evaluated on their goodness of fit to the observed data. The signature differences between models that best fit the passive learning data and models that best fit the active learning data concerned a delayed versus immediate learning parameter and a degree-of-match parameter that determined which patterns were retrieved to make category decisions. A functional account of these parameters is given by considering the learning task as a search process and the role of these parameters in localizing the impact of learning mechanisms in certain areas of the search space. Issues related to simulation as an empirical method for identifying candidate process models are discussed. (shrink)