• Publications
  • Influence
A Model of Inductive Bias Learning
  • J. Baxter
  • Computer Science
  • J. Artif. Intell. Res.
  • 1 February 2000
TLDR
A major problem in machine learning is that of inductive bias: how to choose a learner's hypothesis space so that it is large enough to contain a solution to the problem being learnt, yet small enough to ensure reliable generalization from reasonably-sized training sets. Expand
  • 773
  • 78
  • PDF
Boosting Algorithms as Gradient Descent
TLDR
We provide an abstract characterization of boosting algorithms as gradient decsent on cost-functionals in an inner-product function space. Expand
  • 599
  • 45
  • PDF
Learning internal representations
  • J. Baxter
  • Computer Science, Mathematics
  • COLT '95
  • 5 July 1995
TLDR
We introduce a mechanism for automatically learning an appropriate internal representation for a learning environment and then using that representation to bias the learner's hypothesis space for the learning of future tasks drawn from the same environment. Expand
  • 312
  • 30
  • PDF
A Bayesian/Information Theoretic Model of Learning to Learn via Multiple Task Sampling
  • J. Baxter
  • Computer Science
  • Machine Learning
  • 1 July 1997
TLDR
A Bayesian model of learning to learn by sampling from multiple tasks is presented. Expand
  • 318
  • 19
  • PDF
Variance Reduction Techniques for Gradient Estimates in Reinforcement Learning
TLDR
We study two commonly used policy gradient techniques, the baseline and actor-critic methods, and show that using the average reward to define the baseline can be suboptimal. Expand
  • 324
  • 19
  • PDF
Direct Gradient-Based Reinforcement Learning: I. Gradient Estimation Algorithms
Despite their many empirical successes, approximate value -function based approaches to reinforcement learning suffer from a paucity of theoretical guarantees on the performance of the policyExpand
  • 73
  • 12
  • PDF
Dynamic repositioning of genes in the nucleus of lymphocytes preparing for cell division.
We show that several transcriptionally inactive genes localize to centromeric heterochromatin in the nucleus of cycling but not quiescent (noncycling) primary B lymphocytes. In quiescent cells,Expand
  • 427
  • 11
Topoisomerase II inactivation prevents the completion of DNA replication in budding yeast.
Type II topoisomerases are essential for resolving topologically entwined double-stranded DNA. Although anti-topoisomerase 2 (Top2) drugs are clinically important antibiotics and chemotherapies, toExpand
  • 139
  • 10
Learning to Play Chess Using Temporal Differences
TLDR
In this paper we present TDLEAF(λ), a variation on the TD(λ) algorithm that enables it to be used in conjunction with game-tree search. Expand
  • 119
  • 9
  • PDF