Timezone: »

Q-learning with Nearest Neighbors
Devavrat Shah · Qiaomin Xie

Wed Dec 05 02:00 PM -- 04:00 PM (PST) @ Room 517 AB #119
We consider model-free reinforcement learning for infinite-horizon discounted Markov Decision Processes (MDPs) with a continuous state space and unknown transition kernel, when only a single sample path under an arbitrary policy of the system is available. We consider the Nearest Neighbor Q-Learning (NNQL) algorithm to learn the optimal Q function using nearest neighbor regression method. As the main contribution, we provide tight finite sample analysis of the convergence rate. In particular, for MDPs with a $d$-dimensional state space and the discounted factor $\gamma \in (0,1)$, given an arbitrary sample path with ``covering time'' $L$, we establish that the algorithm is guaranteed to output an $\varepsilon$-accurate estimate of the optimal Q-function using $\Ot(L/(\varepsilon^3(1-\gamma)^7))$ samples. For instance, for a well-behaved MDP, the covering time of the sample path under the purely random policy scales as $\Ot(1/\varepsilon^d),$ so the sample complexity scales as $\Ot(1/\varepsilon^{d+3}).$ Indeed, we establish a lower bound that argues that the dependence of $ \Omegat(1/\varepsilon^{d+2})$ is necessary.

Author Information

Devavrat Shah (Massachusetts Institute of Technology)

Devavrat Shah is a professor of Electrical Engineering & Computer Science and Director of Statistics and Data Science at MIT. He received PhD in Computer Science from Stanford. He received Erlang Prize from Applied Probability Society of INFORMS in 2010 and NeuIPS best paper award in 2008.

Qiaomin Xie (Massachusetts Institute of Technology)

More from the Same Authors