# Classification-based Approximate Policy Iteration: Experiments and Extended Discussions

arXiv: https://arxiv.org/abs/1407.0449 · HTML: https://mlsanity.com/paper/1407.0449/

## Key facts

- Cited by 3 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 2 Jul 2014.

## Built on by (most cited first, 3 of 3)

- [Provably Good Batch Reinforcement Learning Without Great Exploration](https://mlsanity.com/paper/2007.08202/) (2020)
- [Greedification Operators for Policy Optimization: Investigating Forward and Reverse KL Divergences](https://mlsanity.com/paper/2107.08285/) (2021)
- [Fast Convergence of Policy Regret in Learning Stochastic Optimal Control](https://mlsanity.com/paper/2605.26361/) (2026)

## Abstract (arXiv)

Tackling large approximate dynamic programming or reinforcement learning problems requires methods that can exploit regularities, or intrinsic structure, of the problem in hand. Most current methods are geared towards exploiting the regularities of either the value function or the policy. We introduce a general classification-based approximate policy iteration (CAPI) framework, which encompasses a large class of algorithms that can exploit regularities of both the value function and the policy space, depending on what is advantageous. This framework has two main components: a generic value function estimator and a classifier that learns a policy based on the estimated value function. We establish theoretical guarantees for the sample complexity of CAPI-style algorithms, which allow the policy evaluation step to be performed by a wide variety of algorithms (including temporal-difference-style methods), and can handle nonparametric representations of policies. Our bounds on the estimation error of the performance loss are tighter than existing results. We also illustrate this approach empirically on several problems, including a large HIV control task.