This essay gives advice to authors of papers on machine learning, although much of it carries over to other computational disciplines. The issues covered include the material that should appear in a well-balanced paper, factors that arise in different approaches to evaluation, and ways to improve a submission's ability to communicate ideas to its readers. 1. Introduction Although machine learning has become a scientific discipline, the effective communication of its ideas remains an art. Nevertheless, there exist rules of thumb even for practicing art, and in this essay we present some heuristics that we maintain can help machine learning authors improve their papers. Much of this advice applies equally well to other branches of artificial intelligence and even to scientific fields in general, but we will cast it in terms specific to our discipline. Each section addresses a different facet of publications on machine learning. We first address the content appropriate for paper...
Stored only when both papers are in the corpus — 137 citations exist upstream.