skip to main content


Title: Optimistic Initialization for Exploration in Continuous Control
Optimistic initialization underpins many theoretically sound exploration schemes in tabular domains; however, in the deep function approximation setting, optimism can quickly disappear if initialized naively. We propose a framework for more effectively incorporating optimistic initialization into reinforcement learning for continuous control. Our approach uses metric information about the state-action space to estimate which transitions are still unexplored, and explicitly maintains the initial Q-value optimism for the corresponding state-action pairs. We also develop methods for efficiently approximating these training objectives, and for incorporating domain knowledge into the optimistic envelope to improve sample efficiency. We empirically evaluate these approaches on a variety of hard exploration problems in continuous control, where our method outperforms existing exploration techniques.  more » « less
Award ID(s):
1844960 1955361 1717569
NSF-PAR ID:
10321086
Author(s) / Creator(s):
; ; ; ;
Date Published:
Journal Name:
Proceedings of the Thirty-Sixth AAAI Conference on Artificial Intelligence
Format(s):
Medium: X
Sponsoring Org:
National Science Foundation
More Like this
  1. Safe reinforcement learning is extremely challenging--not only must the agent explore an unknown environment, it must do so while ensuring no safety constraint violations. We formulate this safe reinforcement learning (RL) problem using the framework of a finite-horizon Constrained Markov Decision Process (CMDP) with an unknown transition probability function, where we model the safety requirements as constraints on the expected cumulative costs that must be satisfied during all episodes of learning. We propose a model-based safe RL algorithm that we call Doubly Optimistic and Pessimistic Exploration (DOPE), and show that it achieves an objective regret $\tilde{O}(|\mathcal{S}|\sqrt{|\mathcal{A}| K})$ without violating the safety constraints during learning, where $|\mathcal{S}|$ is the number of states, $|\mathcal{A}|$ is the number of actions, and $K$ is the number of learning episodes. Our key idea is to combine a reward bonus for exploration (optimism) with a conservative constraint (pessimism), in addition to the standard optimistic model-based exploration. DOPE is not only able to improve the objective regret bound, but also shows a significant empirical performance improvement as compared to earlier optimism-pessimism approaches. 
    more » « less
  2. null (Ed.)
    Monte-Carlo planning, as exemplified by Monte-Carlo Tree Search (MCTS), has demonstrated remarkable performance in applications with finite spaces. In this paper, we consider Monte-Carlo planning in an environment with continuous state-action spaces, a much less understood problem with important applications in control and robotics. We introduce POLY-HOOT , an algorithm that augments MCTS with a continuous armed bandit strategy named Hierarchical Optimistic Optimization (HOO) (Bubeck et al., 2011). Specifically, we enhance HOO by using an appropriate polynomial, rather than logarithmic, bonus term in the upper confidence bounds. Such a polynomial bonus is motivated by its empirical successes in AlphaGo Zero (Silver et al., 2017b), as well as its significant role in achieving theoretical guarantees of finite space MCTS (Shah et al., 2019). We investigate, for the first time, the regret of the enhanced HOO algorithm in non-stationary bandit problems. Using this result as a building block, we establish non-asymptotic convergence guarantees for POLY-HOOT : the value estimate converges to an arbitrarily small neighborhood of the optimal value function at a polynomial rate. We further provide experimental results that corroborate our theoretical findings. 
    more » « less
  3. We develop provably efficient reinforcement learning algorithms for two-player zero-sum finite-horizon Markov games with simultaneous moves. To incorporate function approximation, we consider a family of Markov games where the reward function and transition kernel possess a linear structure. Both the offline and online settings of the problems are considered. In the offline setting, we control both players and aim to find the Nash equilibrium by minimizing the duality gap. In the online setting, we control a single player playing against an arbitrary opponent and aim to minimize the regret. For both settings, we propose an optimistic variant of the least-squares minimax value iteration algorithm. We show that our algorithm is computationally efficient and provably achieves an [Formula: see text] upper bound on the duality gap and regret, where d is the linear dimension, H the horizon and T the total number of timesteps. Our results do not require additional assumptions on the sampling model. Our setting requires overcoming several new challenges that are absent in Markov decision processes or turn-based Markov games. In particular, to achieve optimism with simultaneous moves, we construct both upper and lower confidence bounds of the value function, and then compute the optimistic policy by solving a general-sum matrix game with these bounds as the payoff matrices. As finding the Nash equilibrium of a general-sum game is computationally hard, our algorithm instead solves for a coarse correlated equilibrium (CCE), which can be obtained efficiently. To our best knowledge, such a CCE-based scheme for optimism has not appeared in the literature and might be of interest in its own right. 
    more » « less
  4. Abstract

    We consider the problem of a leader who can assign rewards for citizens for different anti-regime actions. Citizens face a coordination problem in which each citizen has a private, endogenous degree of optimism about the likelihood of regime change. Because more optimistic citizens are easier to motivate, the choice of optimal rewards entails optimal screening. This leads to a distribution of anti-regime actions. A key result is the emergence of a vanguard, consisting of citizens who engage in the endogenous, maximum level of action. Other citizens participate at varying degrees, with less optimistic citizens contributing less. We explore how the regime’s strength or the maximum reward available to the leader influences the distribution of actions. Moreover, we show that more heterogeneity (e.g., higher inequality) among potential revolutionaries reduces the likelihood of regime change. Our methodological contribution is that we deliver a sharp and novel marriage of screening and global games.

     
    more » « less
  5. We study algorithms using randomized value functions for exploration in reinforcement learning. This type of algorithms enjoys appealing empirical performance. We show that when we use 1) a single random seed in each episode, and 2) a Bernstein-type magnitude of noise, we obtain a worst-case O~(H√SAT) regret bound for episodic time-inhomogeneous Markov Decision Process where S is the size of state space, A is the size of action space, H is the planning horizon and T is the number of interactions. This bound polynomially improves all existing bounds for algorithms based on randomized value functions, and for the first time, matches the Ω(H√SAT) lower bound up to logarithmic factors. Our result highlights that randomized exploration can be near-optimal, which was previously achieved only by optimistic algorithms. To achieve the desired result, we develop 1) a new clipping operation to ensure both the probability of being optimistic and the probability of being pessimistic are lower bounded by a constant, and 2) a new recursive formula for the absolute value of estimation errors to analyze the regret. 
    more » « less