vix.ing · top · new · best · stats · spec

Turing-Universal Learners with Optimal Scaling Laws

2021/11/09 by Preetum Nakkiran, Nakkiran, Preetum · 1 citation
Computer Science · #Algorithms and Data Compression #Artificial Intelligence (cs.AI) #Computability, Logic, AI Algorithms #Computational Complexity (cs.CC) #FOS: Computer and information sciences #FOS: Mathematics #Machine Learning (cs.LG) #Machine Learning (stat.ML) #Machine Learning and Algorithms #Statistics Theory (math.ST)

paper · pdf · doi:10.48550/arxiv.2111.05321

openalex publication_date 2021/11/09 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

For a given distribution, learning algorithm, and performance metric, the rate of convergence (or data-scaling law) is the asymptotic behavior of the algorithm's test performance as a function of number of train samples. Many learning methods in both theory and practice have power-law rates, i.e. performance scales as n for some α> 0. Moreover, both theoreticians and practitioners are concerned with improving the rates of their learning algorithms under settings of interest. We observe the existence of a "universal learner", which achieves the best possible distribution-dependent asymptotic rate among all learning algorithms within a specified runtime (e.g. O(n2)), while incurring only polylogarithmic slowdown over this runtime. This algorithm is uniform, and does not depend on the distribution, and yet achieves best-possible rates for all distributions. The construction itself is a simple extension of Levin's universal search (Levin, 1973). And much like universal search, the universal learner is not at all practical, and is primarily of theoretical and philosophical interest.

Citations

Cited by

Related