A New Model of Computation for Learning Vision Modules from Examples

• Authors: Rhys A. Newman
• Venue: Journal of Mathematical Imaging and Vision
• Year: 1999

This paper addresses an important class of mimicryproblems, where the goal is to construct a computer program which isfunctionally equivalent to an observed behaviour. Computer visionresearch can be considered such a challenge, where a researcherattempts to impart human visual abilities to a computer.Unfortunately this has proved a difficult task, not least because ourvision processes occur mostly at a subconscious level. It istherefore useful to study the general mimicry problem in order todevelop tools which may assist computer vision research.This paper formalises a mimicry problem as one in which a computer learning system (L) constructs a solution from a given program structure (i.e. template or outline) by posing questions to an Oracle. The latter is an entity which, when given an input value, produces the corresponding output of the function which is to be mimicked.In order to define a program‘s structure, particularly one which canbe extracted from any computer program automatically, a new model ofcomputation is developed. Based on this a fast algorithm whichdetermines the best questions to pose to the Oracle is thendescribed. Thus L relieves the human programmer of thedifficulties faced in choosing the examples from which to learn. Thisis important because a human programmer might inadvertently choosebiased, redundant or otherwise unhelpful examples. Results are shownwhich demonstrate the utility of a complete learning system (L) based on this work.This paper represents background theory and initial algorithms which further work will extend into powerful automatic learning systems, examples of which are found in [36] and [38].