Approximate Description Length, Covering Numbers, and VC Dimension
This work clarifies theoretical connections in machine learning complexity theory, but it is incremental as it builds on prior definitions without new empirical results.
The paper investigates the relationship between Approximate Description Length (ADL) and classical complexity measures like Covering Numbers and VC Dimension, finding equivalence for real-valued functions but not for high-dimensional ranges.
Recently, Daniely and Granot [arXiv:1910.05697] introduced a new notion of complexity called Approximate Description Length (ADL). They used it to derive novel generalization bounds for neural networks, that despite substantial work, were out of reach for more classical techniques such as discretization, Covering Numbers and Rademacher Complexity. In this paper we explore how ADL relates to classical notions of function complexity such as Covering Numbers and VC Dimension. We find that for functions whose range is the reals, ADL is essentially equivalent to these classical complexity measures. However, this equivalence breaks for functions with high dimensional range.