The Strength of Weak Learnability

Abstract

The problem considered is that of improving the accuracy of an hypothesis output by a learning algorithm in the distribution-free (PAC) learning model. A concept class is learnable (or strongly learnable) if, given access to a source of examples from the unknown concept, the learner with high probability is able to output an hypothesis that is correct on all but an arbitrarily small fraction of the instances. The concept class is weakly learnable if the learner can produce an hypothesis that performs only slightly better than random guessing. In this paper, it is shown that these two notions of learnability are equivalent. A method is described for converting a weak learning algorithm into one the achieves arbitrarily high accuracy. This construction may have practical applications as a tool for efficiently converting a mediocre learning algorithm into one that performs extremely well. In addition, the construction has some interesting theoretical consequences, including a set of general upper bounds on the complexity of any strong learning algorithm as a function of the allowed error. Keywords: Machine learning, Learning from examples, polynomial-time identification.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Oct 16, 1989
Accession Number
ADA216389

Entities

People

  • Robert E. Schapire

Organizations

  • Massachusetts Institute of Technology

Tags

Communities of Interest

  • Air Platforms
  • Autonomy
  • C4I
  • Human Systems

DTIC Thesaurus Topics

  • Accuracy
  • Algorithms
  • Artificial Intelligence
  • Computational Complexity
  • Computations
  • Computer Science
  • Data Compression
  • Errors
  • Identification
  • Information Systems
  • Language
  • Machine Learning
  • Notation
  • Probability
  • Random Variables
  • Security
  • Symbols

Fields of Study

  • Computer science

Readers

  • Mathematical Modeling and Probability Theory.
  • Neural Network Machine Learning.
  • Regression Analysis.

Technology Areas

  • AI & ML
  • AI & ML - Bayesian Inference
  • AI & ML - Machine Learning Algorithms
  • AI & ML - Neural Networks