Credit Assignment Method for Learning Effective Stochastic Policies in Uncertain Domains (original) (raw)
Related papers
Proceedings of the 1998 Conference on Advances in Neural Information Processing Systems Ii, 1999
Agents acting in the real world are confronted with the problem of making good decisions with limited knowledge of the environment. Partially observable Markov decision processes (POMDPs) model decision problems in which an agent tries to maximize its reward in the face of limited sensor feedback. Recent work has shown empirically that a reinforcement learning (RL) algorithm called Sarsa(A) can efficiently find optimal memoryless policies, which map current observations to actions, for POMDP problems (Loch and Singh 1998). The Sarsa(A) algorithm uses a form of short-term memory called an eligibility trace, which distributes temporally delayed rewards to observation-action pairs which lead up to the reward. This paper explores the effect of eligibility traces on the ability of the Sarsa(A) algorithm to find optimal memoryless policies. A variant of Sarsa(A) called k-step truncated Sarsa(A) is applied to four test problems taken from the recent work of Littman, Littman, Cassandra and Kaelbling, Parr and Russell, and Chrisman. The empirical results show that eligibility traces can be significantly truncated without affecting the ability of Sarsa(A) to find optimal memoryless policies for POMDPs.
1998
Recent research on hidden-state reinforcement learning (RL) problems has concentrated on overcoming partial observability by using memory to estimate state. However, such methods are computationally extremely expensive and thus have very limited applicability. This emphasis on state estimation has come about because it has been widely observed that the presence of hidden state or partial observability renders popular RL methods such as Q-learning and Sarsa useless. However, this observation is misleading in two ways: first, the theoretical results supporting it only apply to RL algorithms that do not use eligibility traces, and second these results are worst-case results, which leaves open the possibility that there may be large classes of hidden-state problems in which RL algorithms work well without any state estimation. In this paper we show empirically that Sarsa(λ), a well known family of RL algorithms that use eligibility traces, can work very well on hidden state problems tha...
Probabilistic Policy Reuse In a Reinforcement Learning Agent
Proceedings of the fifth international joint …, 2006
We contribute Policy Reuse as a technique to improve a reinforcement learning agent with guidance from past learned similar policies. Our method relies on using the past policies as a probabilistic bias where the learning agent faces three choices: the exploitation of the ongoing learned policy, the exploration of random unexplored actions, and the exploitation of past policies. We introduce the algorithm and its major components: an exploration strategy to include the new reuse bias, and a similarity function to estimate the similarity of past policies with respect to a new one. We provide empirical results demonstrating that Policy Reuse improves the learning performance over different strategies that learn without reuse. Interestingly and almost as a side effect, Policy Reuse also identifies classes of similar policies revealing a basis of core policies of the domain. We demonstrate that such a basis can be built incrementally, contributing the learning of the structure of a domain.
Learning without state-estimation in partially observable Markovian decision problems
Icml, 1984
Reinforcement learning (RL) algorithms provide a sound theoretical basis for building learning control architectures for embedded agents. Unfortunately all of the theory and much of the practice (see for an exception) of RL is limited to Markovian decision processes (MDPs). Many realworld decision tasks, however, are inherently non-Markovian, i.e., the state of the environment is only incompletely known to the learning agent. In this paper we consider only partially observable MDPs (POMDPs), a useful class of non-Markovian decision processes. Most previous approaches to such problems have combined computationally expensive state-estimation techniques with learning control. This paper investigates learning in POMDPs without resorting to any f o r m of state estimation. We present results about what TD(0) and Q-learning will do when applied to POMDPs. It is shown that the conventional discounted RL framework is inadequate to deal with POMDPs. Finally we develop a new framework for learning without state-estimation in POMDPs by including stochastic policies in the search space, and by de ning the value or utility o f a distribution over states.
Using Rewards for Belief State Updates in Partially Observable Markov Decision Processes
Lecture Notes in Computer Science, 2005
Partially Observable Markov Decision Processes (POMDP) provide a standard framework for sequential decision making in stochastic environments. In this setting, an agent takes actions and receives observations and rewards from the environment. Many POMDP solution methods are based on computing a belief state, which is a probability distribution over possible states in which the agent could be. The action choice of the agent is then based on the belief state. The belief state is computed based on a model of the environment, and the history of actions and observations seen by the agent. However, reward information is not taken into account in updating the belief state. In this paper, we argue that rewards can carry useful information that can help disambiguate the hidden state. We present a method for updating the belief state which takes rewards into account. We present experiments with exact and approximate planning methods on several standard POMDP domains, using this belief update method, and show that it can provide advantages, both in terms of speed and in terms of the quality of the solution obtained.
Learning policies with external memory
2001
In order for an agent to perform well in partially observable domains, it is usually necessary for actions to depend on the history of observations. In this paper, we explore a stigmergic approach, in which the agent's actions include the ability to set and clear bits in an external memory, and the external memory is included as part of the input to the agent. In this case, we need to learn a reactive policy in a highly non-Markovian domain. We explore two algorithms: sarsa(λ), which has had empirical success in partially observable domains, and vaps, a new algorithm due to Baird and Moore, with convergence guarantees in partially observable domains. We compare the performance of these two algorithms on benchmark problems.
Stochastic abstract policies: generalizing knowledge to improve reinforcement learning
IEEE transactions on cybernetics, 2015
Reinforcement learning (RL) enables an agent to learn behavior by acquiring experience through trial-and-error interactions with a dynamic environment. However, knowledge is usually built from scratch and learning to behave may take a long time. Here, we improve the learning performance by leveraging prior knowledge; that is, the learner shows proper behavior from the beginning of a target task, using the knowledge from a set of known, previously solved, source tasks. In this paper, we argue that building stochastic abstract policies that generalize over past experiences is an effective way to provide such improvement and this generalization outperforms the current practice of using a library of policies. We achieve that contributing with a new algorithm, AbsProb-PI-multiple and a framework for transferring knowledge represented as a stochastic abstract policy in new RL tasks. Stochastic abstract policies offer an effective way to encode knowledge because the abstraction they provid...
Learning in Markov Decision Processes under Constraints
ArXiv, 2020
We consider reinforcement learning (RL) in Markov Decision Processes (MDPs) in which at each time step the agent, in addition to earning a reward, also incurs an MMM dimensional vector of costs. The objective is to design a learning rule that maximizes the cumulative reward earned over a finite time horizon of TTT steps, while simultaneously ensuring that the cumulative cost expenditures are bounded appropriately. The considerations on the cumulative cost expenditures is in departure from the existing RL literature, in that the agent now additionally needs to balance the cost expenses in an \emph{online manner}, while simultaneously performing optimally the exploration-exploitation trade-off typically encountered in RL tasks. This is challenging since either of the duo objectives of exploration and exploitation necessarily require the agent to expend resources. When the constraints are placed on the average costs, we present a version of UCB algorithm and prove that its reward as we...
Learning Probabilistic Reward Machines from Non-Markovian Stochastic Reward Processes
2021
The success of reinforcement learning in typical settings is, in part, predicated on underlying Markovian assumptions on the reward signal by which an agent learns optimal policies. In recent years, the use of reward machines has relaxed this assumption by enabling a structured representation of non-Markovian rewards. In particular, such representations can be used to augment the state space of the underlying decision process, thereby facilitating non-Markovian reinforcement learning. However, these reward machines cannot capture the semantics of stochastic reward signals. In this paper, we make progress on this front by introducing probabilistic reward machines (PRMs) as a representation of non-Markovian stochastic rewards. We present an algorithm to learn PRMs from the underlying decision process as well as to learn the PRM representation of a given decisionmaking policy.