Robust learning aided by context

Empirical st,udies of multitask learning provide some evidence that the performance of a learning system on its intended targets improves by presenting to the learning system related tasks, also called contexts, as additional input. Angluin, Gasarch, and Smith, as well as Kinber, Smith, Velauthapillai, and Wiehagen have provided mathematical justification for this phenomenon in the inductive inference framework. However, their proofs rely heavily on self-referential coding tricks, that is, they directly code the solution of the learning problem into the context. Fulk has shown that for the Exand Bc-anomaly hierarchies. such results, which rely on self-referential coding tricks, may not hold robustly. In this work we analyze robust versions of learning aided by context and show that ~~ in contrast to Fulk’s result above --the rob,ust versions of *This work was carried out while J. Case, S. Jain, M. Ott, and F. Stephan were visiting the School of Computer Science and Engineering at the University of New South Wales. ‘Supported hy the Deutsche Forschungsgemeinschaft (DFG) Graduiertenkolleg *‘Beherrschbarkeit, komplexer Systen:; (GRK 209/Z-96). upported by Australian Research Council Grant A49600456. SSupported by t,he Deutsche Forschungsgemeinschaft (DF’G) Grant Am 60/9-Z. Pennission to make digital or hard copies ofnll or part ofthis work for personal or classroom use is gmnted without fee provided that copies are not made or distributed for profit or commercial adwantage and that copies bear this notice and the full citation on the first page. To copy otherwise, to republish, to post on servem or to redistrihute to lists, requires prior specific permission and/or a fee. COLT 98 Madison WI 1JSA Copyright ACM 1998 l-581 13-057--O/981 7...$5.00 t,hese learning notions are st,ill very powerful. Also, studied is the difficulty of the functional dependence between the intended target, tasks and useful associated contexts.