Current browse context:
stat.ML
Change to browse by:
References & Citations
Statistics > Machine Learning
Title: TopRank+: A Refinement of TopRank Algorithm
(Submitted on 21 Jan 2020)
Abstract: Online learning to rank is a core problem in machine learning. In Lattimore et al. (2018), a novel online learning algorithm was proposed based on topological sorting. In the paper they provided a set of self-normalized inequalities (a) in the algorithm as a criterion in iterations and (b) to provide an upper bound for cumulative regret, which is a measure of algorithm performance. In this work, we utilized method of mixtures and asymptotic expansions of certain implicit function to provide a tighter, iterated-log-like boundary for the inequalities, and as a consequence improve both the algorithm itself as well as its performance estimation.
Link back to: arXiv, form interface, contact.