Showing papers on "Active learning (machine learning) published in 1979"

Models of Learning Systems.

Abstract: : The terms adaptation, learning, concept-formation, induction, self-organization, and self-repair have all been used in the context of learning system (LS) research. In this article, three distinct approaches to machine learning and adaptation are considered: (i) the adaptive control approach, (ii) the pattern recognition approach, and (iii) the artificial intelligence approach. Progress in each of these areas is summarized in the first part of the article. In the next part a general model for learning systems is presented that allows characterization and comparison of individual algorithms and programs in all of these areas. The model details the functional components felt to be essential for any learning system, independent of the techniques used for its construction, and the specific environment in which it operates. Specific examples of learning systems are described in terms of the model. (Author)

Learning under a VEDIC teacher

Abstract: The problem of parametric learning under a vicissitudinous teacher,i.e., a teacher with unknown variable characteristics, is the topic of this study. The concept central to the technique developed here is that learning the variable characteristics of the teacher aids the parametric learning under such vicissitudinous environment. This is demonstrated effectively through presentation of simulation results. It is shown that the efficiency of the parametric learning under randomly varying levels of supervision is significantly enhanced by tracking the variable characteristics of the VEDIC teacher (for each pattern class) during the learning process.

Actively adaptive methods for stochastic systems

Abstract: An important element of adaptive control is learning of the drifting parameters. As the process unfolds, additional information becomes available, which will provide learning for the purpose of control. This information may come about accidentally through past control actions or as a result of active probing, which itself is a possible control policy. Thus learning is present, where it is accidental or deliberate. Since more learning may improve overall control performance, the probing signal may indirectly help in controlling the stochastic system. On the other hand, excessive probing should not be allowed even though it may promote learning because it is expensive in the sense that it will, in general, increase the expected cost performance of the system. A good control law must then regulate its adaptation (learning) in an optimal manner. An adaptive control method is called passively adaptive if learning is not planned in the manner described above; it is called actively adaptive if learning is planned and regulated for the purpose of final control. This paper gives an overview of adaptive control methods which were developed based on the concept of active learning for control purposes. Some comments on their practicality are also given.

Mathematical learning theory: a formalized, axiomatic, abstract approach

Abstract: An abstract formalism is presented wherein a mathematical learning theory is explored Numerous examples from the literature are presented demonstrating how our axiomatic framework formally unifies diverse examples of pattern recognition Our principal result, the adaptive learning theorem , discusses the expected waiting time of different learning machines Roughly speaking it states that all algorithms for the induction of boolean relationships between a dependent binary parameter and a finite preassigned set of independent parameters have asymptotically identical average rates of learning We also give a lower bound for the average learning speed as well as necessary and sufficient conditions for a machine to achieve it We discuss applications of our methods to classical pattern recognition problems as well as possible application to more complicated research problems Our principal mathematical innovation is a fruitful correspondence between probability spaces and labeled tree representations