Get our free extension to see links to code for papers anywhere online!Free add-on: code for papers everywhere!Free add-on: See code for papers anywhere!

Abstract:In reinforcement learning, offline value function learning is the procedure of using an offline dataset to estimate the expected discounted return from each state when taking actions according to a fixed target policy. The stability of this procedure, i.e., whether it converges to its fixed-point, critically depends on the representations of the state-action pairs. Poorly learned representations can make value function learning unstable, or even divergent. Therefore, it is critical to stabilize value function learning by explicitly shaping the state-action representations. Recently, the class of bisimulation-based algorithms have shown promise in shaping representations for control. However, it is still unclear if this class of methods can stabilize value function learning. In this work, we investigate this question and answer it affirmatively. We introduce a bisimulation-based algorithm called kernel representations for offline policy evaluation (KROPE). KROPE uses a kernel to shape state-action representations such that state-action pairs that have similar immediate rewards and lead to similar next state-action pairs under the target policy also have similar representations. We show that KROPE: 1) learns stable representations and 2) leads to lower value error than baselines. Our analysis provides new theoretical insight into the stability properties of bisimulation-based methods and suggests that practitioners can use these methods for stable and accurate evaluation of offline reinforcement learning agents.

Via

Abstract:We present Reinforcement Learning via Auxiliary Task Distillation (AuxDistill), a new method that enables reinforcement learning (RL) to perform long-horizon robot control problems by distilling behaviors from auxiliary RL tasks. AuxDistill achieves this by concurrently carrying out multi-task RL with auxiliary tasks, which are easier to learn and relevant to the main task. A weighted distillation loss transfers behaviors from these auxiliary tasks to solve the main task. We demonstrate that AuxDistill can learn a pixels-to-actions policy for a challenging multi-stage embodied object rearrangement task from the environment reward without demonstrations, a learning curriculum, or pre-trained skills. AuxDistill achieves $2.3 \times$ higher success than the previous state-of-the-art baseline in the Habitat Object Rearrangement benchmark and outperforms methods that use pre-trained skills and expert demonstrations.

Via

Abstract:In this paper, we study multi-task structured bandit problem where the goal is to learn a near-optimal algorithm that minimizes cumulative regret. The tasks share a common structure and the algorithm exploits the shared structure to minimize the cumulative regret for an unseen but related test task. We use a transformer as a decision-making algorithm to learn this shared structure so as to generalize to the test task. The prior work of pretrained decision transformers like DPT requires access to the optimal action during training which may be hard in several scenarios. Diverging from these works, our learning algorithm does not need the knowledge of optimal action per task during training but predicts a reward vector for each of the actions using only the observed offline data from the diverse training tasks. Finally, during inference time, it selects action using the reward predictions employing various exploration strategies in-context for an unseen test task. Our model outperforms other SOTA methods like DPT, and Algorithmic Distillation over a series of experiments on several structured bandit problems (linear, bilinear, latent, non-linear). Interestingly, we show that our algorithm, without the knowledge of the underlying problem structure, can learn a near-optimal policy in-context by leveraging the shared structure across diverse tasks. We further extend the field of pre-trained decision transformers by showing that they can leverage unseen tasks with new actions and still learn the underlying latent structure to derive a near-optimal policy. We validate this over several experiments to show that our proposed solution is very general and has wide applications to potentially emergent online and offline strategies at test time. Finally, we theoretically analyze the performance of our algorithm and obtain generalization bounds in the in-context multi-task learning setting.

Via

Abstract:In this paper, we study safe data collection for the purpose of policy evaluation in tabular Markov decision processes (MDPs). In policy evaluation, we are given a \textit{target} policy and asked to estimate the expected cumulative reward it will obtain. Policy evaluation requires data and we are interested in the question of what \textit{behavior} policy should collect the data for the most accurate evaluation of the target policy. While prior work has considered behavior policy selection, in this paper, we additionally consider a safety constraint on the behavior policy. Namely, we assume there exists a known default policy that incurs a particular expected cost when run and we enforce that the cumulative cost of all behavior policies ran is better than a constant factor of the cost that would be incurred had we always run the default policy. We first show that there exists a class of intractable MDPs where no safe oracle algorithm with knowledge about problem parameters can efficiently collect data and satisfy the safety constraints. We then define the tractability condition for an MDP such that a safe oracle algorithm can efficiently collect data and using that we prove the first lower bound for this setting. We then introduce an algorithm SaVeR for this problem that approximates the safe oracle algorithm and bound the finite-sample mean squared error of the algorithm while ensuring it satisfies the safety constraint. Finally, we show in simulations that SaVeR produces low MSE policy evaluation while satisfying the safety constraint.

Via

Abstract:General Value Functions (GVFs) (Sutton et al, 2011) are an established way to represent predictive knowledge in reinforcement learning. Each GVF computes the expected return for a given policy, based on a unique pseudo-reward. Multiple GVFs can be estimated in parallel using off-policy learning from a single stream of data, often sourced from a fixed behavior policy or pre-collected dataset. This leaves an open question: how can behavior policy be chosen for data-efficient GVF learning? To address this gap, we propose GVFExplorer, which aims at learning a behavior policy that efficiently gathers data for evaluating multiple GVFs in parallel. This behavior policy selects actions in proportion to the total variance in the return across all GVFs, reducing the number of environmental interactions. To enable accurate variance estimation, we use a recently proposed temporal-difference-style variance estimator. We prove that each behavior policy update reduces the mean squared error in the summed predictions over all GVFs. We empirically demonstrate our method's performance in both tabular representations and nonlinear function approximation.

Via

Abstract:On-policy reinforcement learning (RL) algorithms perform policy updates using i.i.d. trajectories collected by the current policy. However, after observing only a finite number of trajectories, on-policy sampling may produce data that fails to match the expected on-policy data distribution. This sampling error leads to noisy updates and data inefficient on-policy learning. Recent work in the policy evaluation setting has shown that non-i.i.d., off-policy sampling can produce data with lower sampling error than on-policy sampling can produce. Motivated by this observation, we introduce an adaptive, off-policy sampling method to improve the data efficiency of on-policy policy gradient algorithms. Our method, Proximal Robust On-Policy Sampling (PROPS), reduces sampling error by collecting data with a behavior policy that increases the probability of sampling actions that are under-sampled with respect to the current policy. Rather than discarding data from old policies -- as is commonly done in on-policy algorithms -- PROPS uses data collection to adjust the distribution of previously collected data to be approximately on-policy. We empirically evaluate PROPS on both continuous-action MuJoCo benchmark tasks as well as discrete-action tasks and demonstrate that (1) PROPS decreases sampling error throughout training and (2) improves the data efficiency of on-policy policy gradient algorithms. Our work improves the RL community's understanding of a nuance in the on-policy vs off-policy dichotomy: on-policy learning requires on-policy data, not on-policy sampling.

Via

Abstract:We study multi-task representation learning for the problem of pure exploration in bilinear bandits. In bilinear bandits, an action takes the form of a pair of arms from two different entity types and the reward is a bilinear function of the known feature vectors of the arms. In the \textit{multi-task bilinear bandit problem}, we aim to find optimal actions for multiple tasks that share a common low-dimensional linear representation. The objective is to leverage this characteristic to expedite the process of identifying the best pair of arms for all tasks. We propose the algorithm GOBLIN that uses an experimental design approach to optimize sample allocations for learning the global representation as well as minimize the number of samples needed to identify the optimal pair of arms in individual tasks. To the best of our knowledge, this is the first study to give sample complexity analysis for pure exploration in bilinear bandits with shared representation. Our results demonstrate that by learning the shared representation across tasks, we achieve significantly improved sample complexity compared to the traditional approach of solving tasks independently.

Via

Abstract:Learning from demonstration (LfD) is a popular technique that uses expert demonstrations to learn robot control policies. However, the difficulty in acquiring expert-quality demonstrations limits the applicability of LfD methods: real-world data collection is often costly, and the quality of the demonstrations depends greatly on the demonstrator's abilities and safety concerns. A number of works have leveraged data augmentation (DA) to inexpensively generate additional demonstration data, but most DA works generate augmented data in a random fashion and ultimately produce highly suboptimal data. In this work, we propose Guided Data Augmentation (GuDA), a human-guided DA framework that generates expert-quality augmented data. The key insight of GuDA is that while it may be difficult to demonstrate the sequence of actions required to produce expert data, a user can often easily identify when an augmented trajectory segment represents task progress. Thus, the user can impose a series of simple rules on the DA process to automatically generate augmented samples that approximate expert behavior. To extract a policy from GuDA, we use off-the-shelf offline reinforcement learning and behavior cloning algorithms. We evaluate GuDA on a physical robot soccer task as well as simulated D4RL navigation tasks, a simulated autonomous driving task, and a simulated soccer task. Empirically, we find that GuDA enables learning from a small set of potentially suboptimal demonstrations and substantially outperforms a DA strategy that samples augmented data randomly.

Via

Abstract:In reinforcement learning, off-policy evaluation (OPE) is the problem of estimating the expected return of an evaluation policy given a fixed dataset that was collected by running one or more different policies. One of the more empirically successful algorithms for OPE has been the fitted q-evaluation (FQE) algorithm that uses temporal difference updates to learn an action-value function, which is then used to estimate the expected return of the evaluation policy. Typically, the original fixed dataset is fed directly into FQE to learn the action-value function of the evaluation policy. Instead, in this paper, we seek to enhance the data-efficiency of FQE by first transforming the fixed dataset using a learned encoder, and then feeding the transformed dataset into FQE. To learn such an encoder, we introduce an OPE-tailored state-action behavioral similarity metric, and use this metric and the fixed dataset to learn an encoder that models this metric. Theoretically, we show that this metric allows us to bound the error in the resulting OPE estimate. Empirically, we show that other state-action similarity metrics lead to representations that cannot represent the action-value function of the evaluation policy, and that our state-action representation method boosts the data-efficiency of FQE and lowers OPE error relative to other OPE-based representation learning methods on challenging OPE tasks. We also empirically show that the learned representations significantly mitigate divergence of FQE under varying distribution shifts. Our code is available here: https://github.com/Badger-RL/ROPE.

Via

Abstract:Recently, data augmentation (DA) has emerged as a method for leveraging domain knowledge to inexpensively generate additional data in reinforcement learning (RL) tasks, often yielding substantial improvements in data efficiency. While prior work has demonstrated the utility of incorporating augmented data directly into model-free RL updates, it is not well-understood when a particular DA strategy will improve data efficiency. In this paper, we seek to identify general aspects of DA responsible for observed learning improvements. Our study focuses on sparse-reward tasks with dynamics-invariant data augmentation functions, serving as an initial step towards a more general understanding of DA and its integration into RL training. Experimentally, we isolate three relevant aspects of DA: state-action coverage, reward density, and the number of augmented transitions generated per update (the augmented replay ratio). From our experiments, we draw two conclusions: (1) increasing state-action coverage often has a much greater impact on data efficiency than increasing reward density, and (2) decreasing the augmented replay ratio substantially improves data efficiency. In fact, certain tasks in our empirical study are solvable only when the replay ratio is sufficiently low.

Via