Random Projection

Practically and provably efficient RL under resource constraints!

First probabilistic framework for sequential random projection, an approach rooted in the challenges of sequential decision-making under uncertainty; A non-trivial martingale extension of Johnson-Lindenstrauss (JL) to sequentially adaptive data processes.

TL;DR: We simplify and unify various constructions of the Johnson-Lindenstrauss (JL) lemma, including spherical and sub-Gaussian models, and provide the first rigorous proof for spherical construction’s effectiveness. Our work extends the Hanson-Wright inequality and solidifies the JL lemma’s theoretical foundation, enhancing its practical applications in computational algorithms.

Practically and provably efficient RL under resource constraints!

Addressing efficiency chanllenge in RL by HyperFQI algorithm