Showing papers on "Unsupervised learning published in 1973"

Pattern classification and scene analysis

Abstract: Provides a unified, comprehensive and up-to-date treatment of both statistical and descriptive methods for pattern recognition. The topics treated include Bayesian decision theory, supervised and unsupervised learning, nonparametric techniques, discriminant analysis, clustering, preprosessing of pictorial data, spatial filtering, shape description techniques, perspective transformations, projective invariants, linguistic procedures, and artificial intelligence techniques for scene analysis.

Punish/Reward: Learning with a Critic in Adaptive Threshold Systems

Abstract: An adaptive threshold element is able to "learn" a strategy of play for the game blackjack (twenty-one) with a performance close to that of the Thorp optimal strategy although the adaptive system has no prior knowledge of the game and of the objective of play. After each winning game the decisions of the adaptive system are "rewarded." After each losing game the decisions are "punished." Reward is accomplished by adapting while accepting the actual decision as the desired response. Punishment is accomplished by adapting while taking the desired response to be the opposite of that of the actual decision. This learning scheme is unlike "learning with a teacher" and unlike "unsupervised learning." It involves "bootstrap adaptation" or "learning with a critic." The critic rewards decisions which are members of successful chains of decisions and punishes other decisions. A general analytical model for learning with a critic is formulated and analyzed. The model represents bootstrap learning per se. Although the hypotheses on which the model is based do not perfectly fit blackjack learning, it is applied heuristically to predict adaptation rates with good experimental success. New applications are being explored for bootstrap learning in adaptive controls and multilayered adaptive systems.

A Continuous Two-Dimensional Model of Threshold Learning

Abstract: We present a model of threshold learning that represents discrete one-dimensional learning processes by a continuous two-dimensional learning process. The model gives us an overall view of the learning dynamics of an expanded range of training procedures, and provides insight into the training of multidimensional linear pattern classifiers. In our model the expected performance is measured by learning curves, and the confidence in this expected performance is measured by variance curves. Previous work on the continuous approximation has been restricted to one-dimensional learning processes. The theory developed here promises to lead to improved training procedures and improved rules for determining when to stop training.

Unsupervised learning of the Kalman filter

Abstract: It is the purpose of the letter to provide a short description of a general recursive method for learning the optimal stationary Kalman filter Kopt, when the plant and measurement noise covariance kernels, denoted by Q and R, respectively, are unknown. Experimental verification that confirms the theoretical expectations is presented.

A Learning Machine with New Structure and Function

Abstract: A new learning model is proposed, which is quite different from the usual ones in both function and structure. In the usual models only the classification function of input patterns is learned. On the other hand, the proposed learning machine can memorize input patterns themselves by a learning process. That is, an input pattern appears on the output plane of the machine if it has been presented to the machine often enough and does not appear if it has been presented infrequently. Since a man can remember or actually sketch patterns, he not only classifies but probably also memorizes patterns themselves by learning. This machine can therefore be said to simulate a phase of brain functioning. As for structure, the new machine consists of an iteration of nonlinear and spatially local processing instead of the usual single processing with linear discriminant functions. Both a general analysis and experimental results are given, and the machine is shown to have the expected behavior.

On the realization of an optimal decision rule by means of a hierarchial system

Abstract: This paper deals with the realization of a best approximation of an unknown optimal decision rule by means of a self-learning hierarchical system. This optimization implies the simultaneous use of adaptive estimation and classification techniques. An unsupervised learning discrete algorithm based on stochastic approximation theory is proposed as a solution to this problem. Some applications are suggested, particularly, in the field of modelling of static and dynamic non-linear relationships. Numerical results are given.

Sequential non-Gaussian pattern recognition with supervised learning

Abstract: This paper considers binary pattern recognition of a non-Gaussian pattern in Gaussian noise using supervised learning. The scheme is both structure and parameter adaptive. To facilitate a feasible solution, certain judicious approximations are used. Two examples are presented to demonstrate the learning capability of the proposed algorithms.