Nonuniform learnability

Grammatical Inference: Algorithms and …, 2000

The PAC and other equivalent learning models are widely accepted models for polynomial learnability of concept classes. However, negative results abound in the PAC learning framework (concept classes such as deterministic finite state automata (DFA) are not efficiently learnable in the PAC model). The PAC model's requirement of learnability under all conceivable distributions could be considered too stringent a restriction for practical applications. Several models for learning in more helpful environments have been proposed in the literature including: learning from example based queries [2], online learning allowing a bounded number of mistakes , learning with the help of teaching sets , learning from characteristic sets , and learning from simple examples . Several concept classes that are not learnable in the standard PAC model have been shown to be learnable in these models. In this paper we identify the relationships between these different learning models. We also address the issue of unnatural collusion between the teacher and the learner that can potentially trivialize the task of learning in helpful environments.

2020

An earlier introduced characterization of nonuniform learnability that allows the sample size to depend on the hypothesis to which the learner is compared has been redefined using the measure theoretic approach. Where nonuniform learnability is a strict relaxation of the Probably Approximately Correct framework. Introduction of a new algorithm, Generalize Measure Learnability framework, to implement this approach with the study of its sample and computational complexity bounds. Like the Minimum Description Length principle, this approach can be regarded as an explication of Occam razor. Furthermore, many situations were presented, Hypothesis Classes that are countable where we can apply the GML framework, which we can learn to use the GML scheme and can achieve statistical consistency.

arXiv (Cornell University), 2022

Journal of The ACM, 1989

Valiant's learnability model is extended to learning classes of concepts defined by regions in Euclidean space E". The methods in this paper lead to a unified treatment of some of Valiant's results, along with previous results on distribution-free convergence of certain pattern recognition algorithms. It is shown that the essential condition for distribution-free learnability is finiteness of the Vapnik-Chervonenkis dimension, a simple combinatorial parameter of the class of concepts to be learned. Using this parameter, the complexity and closure properties of learnable classes are analyzed, and the necessary and sufftcient conditions are provided for feasible learnability.

Information and Computation, 1999

Important refinements of concept learning in the limit from positive data considerably restricting the accessibility of input data are studied. Let c be any concept; every infinite sequence of elements exhausting c is called positive presentation of c. In all learning models considered the learning machine computes a sequence of hypotheses about the target concept from a positive presentation of it. With iterative learning, the learning machine, in making a conjecture, has access to its previous conjecture and the latest data item coming in. In k-bounded example-memory inference (k is a priori fixed) the learner is allowed to access, in making a conjecture, its previous hypothesis, its memory of up to k data items it has already seen, and the next element coming in. In the case of k-feedback identification, the learning machine, in making a conjecture, has access to its previous conjecture, the latest data item coming in, and, on the basis of this information, it can compute k items and query the database of previous data to find out, for each of the k items, whether or not it is in the database (k is again a priori fixed). In all cases, the sequence of conjectures has to converge to a hypothesis correctly describing the target concept. Our results are manyfold. An infinite hierarchy of more and more powerful feedback learners in dependence on the number k of queries allowed to be asked is established. However, the hierarchy collapses to 1-feedback inference if only indexed families of infinite concepts are considered, and moreover, its learning power is then equal to learning in the limit. But it remains infinite for concept classes of only infinite r.e. concepts. Both k-feedback inference and k-bounded example-memory identification are more powerful than iterative learning but incomparable to one another. Furthermore, there are cases where redundancy in the hypothesis space is shown to be a resource increasing the learning power of iterative learners. Finally, the union of at most k pattern languages is shown to be iteratively inferable.

Theoretical Computer Science

Measure and category (or rather, their recursion-theoretical counterparts) have been used in theoretical computer science to make precise the intuitive notion "for most of the recursive sets". We use the notions of effective measure and category to discuss the relative sizes of inferrible sets, and their complements.

Journal of Computer and System Sciences, 2005

We study a model of Probably Exactly Correct (PExact) learning that can be viewed either as the Exact model (learning from Equivalence Queries only) relaxed so that counterexamples to equivalence queries are distributionally drawn rather than adversarially chosen or as the Probably Approximately Correct (PAC) model strengthened to require a perfect hypothesis. We also introduce a model of Probably Almost Exactly Correct (PAExact) learning that requires a hypothesis with negligible error and thus lies between the PExact and PAC models. Unlike the Exact and PExact models, PAExact learning is applicable to classes of functions defined over infinite instance spaces. We obtain a number of separation results between these models. Of particular note are some positive results for efficient parallel learning in the PAExact model, which stand in stark contrast to earlier negative results for efficient parallel Exact learning.

Theoretical Computer Science, 2013

1994

The fundamental tradeoff that is well known in Knowledge Representation and Reasoning affects Concept Learning from Examples too. Representation of learning examples using attribute-value has proved to support efficient inductive algorithms but limited expressiveness whereas more expressive representation languages, typically subsets of First Order Logic (FOL), are supported by less efficient algorithms. In fact, an underlying problem is that of the number of different ways of matching examples, just one in attribute-value representation and potentially large in FOL representation. This paper describes a novel approach to perform representation shifts on learning examples. The structure of these learning examples, initially represented using a subset of FOL-based languages, is reformulated so as to produce new learning examples that are represented using an attribute-value language. What is considered to be an adequate structure varies according to the learning task. We introduce the notion of morion (from the Greek) to qualify this structure and show, through a concrete example, the advantages it offers. We then describe an algorithm which reformulates learning examples automatically and go on to analyze its complexity. This approach to deductive reformulation is implemented in the REMO system that has been experimented on the learning of the construction of Chinese characters.

Proceedings of the ninth annual conference on Computational learning theory - COLT '96, 1996

In the k-Restricted-Focus-of-Attention (k-RFA) model, only k of the n attributes of each example are revealed to the learner, although the set of visible attributes in each example is determined by the learner. While the k-RFA model is a natural extension of the PAC model, there are also significant differences. For example, it was previously known that learnability in this model is not characterized by the VC-dimension and that many PAC learning algorithms are not applicable in the k-RFA setting.