skip to main content

Explore Research Products in the NSF-PAR

Title: Autonomous Learning in a Pseudo-Episodic Physical Environment
Abstract Forpractical considerations reinforcement learning has proven to be a difficult task outside of simulation when applied to a physical experiment. Here we derive an optional approach to model free reinforcement learning, achieved entirely online, through careful experimental design and algorithmic decision making. We design a reinforcement learning scheme to implement traditionally episodic algorithms for an unstable 1-dimensional mechanical environment. The training scheme is completely autonomous, requiring no human to be present throughout the learning process. We show that the pseudo-episodic technique allows for additional learning updates with off-policy actor-critic and experience replay methods. We show that including these additional updates between periods of traditional training episodes can improve speed and consistency of learning. Furthermore, we validate the procedure in experimental hardware. In the physical environment, several algorithm variants learned rapidly, each surpassing baseline maximum reward. The algorithms in this research are model free and use only information obtained by an onboard sensor during training.  more » « less
Award ID(s):
1935216
NSF-PAR ID:
10378241
Author(s) / Creator(s):
;
Date Published:
Journal Name:
Journal of Intelligent & Robotic Systems
Volume:
104
Issue:
2
ISSN:
0921-0296
Format(s):
Medium: X
Sponsoring Org:
National Science Foundation
More Like this
  1. ; ; ; ; ; ( , Proc. of the 21st International Conference on Autonomous Agents and Multiagent Systems (AAMAS 2022),)
    Piotr Faliszewski ; Viviana Mascardi (Ed.)
    Recent success in reinforcement learning (RL) has brought renewed attention to the design of reward functions by which agent behavior is reinforced or deterred. Manually designing reward functions is tedious and error-prone. An alternative approach is to specify a formal, unambiguous logic requirement, which is automatically translated into a reward function to be learned from. Omega-regular languages, of which Linear Temporal Logic (LTL) is a subset, are a natural choice for specifying such requirements due to their use in verification and synthesis. However, current techniques based on omega-regular languages learn in an episodic manner whereby the environment is periodically reset to an initial state during learning. In some settings, this assumption is challenging or impossible to satisfy. Instead, in the continuing setting the agent explores the environment without resets over a single lifetime. This is a more natural setting for reasoning about omega-regular specifications defined over infinite traces of agent behavior. Optimizing the average reward instead of the usual discounted reward is more natural in this case due to the infinite-horizon objective that poses challenges to the convergence of discounted RL solutions. We restrict our attention to the omega-regular languages which correspond to absolute liveness specifications. These specifications cannot be invalidated by any finite prefix of agent behavior, in accordance with the spirit of a continuing problem. We propose a translation from absolute liveness omega-regular languages to an average reward objective for RL. Our reduction can be done on-the-fly, without full knowledge of the environment, thereby enabling the use of model-free RL algorithms. Additionally, we propose a reward structure that enables RL without episodic resetting in communicating MDPs, unlike previous approaches. We demonstrate empirically with various benchmarks that our proposed method of using average reward RL for continuing tasks defined by omega-regular specifications is more effective than competing approaches that leverage discounted RL. 
    more » « less
  2. ; ( , Conference on Learning Theory)
    We study model-free reinforcement learning (RL) algorithms for infinite-horizon average-reward Markov decision process (MDP), which is more appropriate for applications that involve continuing operations not divided into episodes. In contrast to episodic/discounted MDPs, theoretical understanding of model-free RL algorithms is relatively inadequate for the average-reward setting. In this paper, we consider both the online setting and the setting with access to a simulator. We develop computationally efficient model-free algorithms that achieve sharper guarantees on regret/sample complexity compared with existing results. In the online setting, we design an algorithm, UCB-AVG, based on an optimistic variant of variance-reduced Q-learning. We show that UCB-AVG achieves a regret bound $\widetilde{O}(S^5A^2sp(h^*)\sqrt{T})$ after $T$ steps, where $S\times A$ is the size of state-action space, and $sp(h^*)$ the span of the optimal bias function. Our result provides the first computationally efficient model-free algorithm that achieves the optimal dependence in $T$ (up to log factors) for weakly communicating MDPs, which is necessary for low regret. In contrast, prior results either are suboptimal in $T$ or require strong assumptions of ergodicity or uniformly mixing of MDPs. In the simulator setting, we adapt the idea of UCB-AVG to develop a model-free algorithm that finds an $\epsilon$-optimal policy with sample complexity $\widetilde{O}(SAsp^2(h^*)\epsilon^{-2} + S^2Asp(h^*)\epsilon^{-1}).$ This sample complexity is near-optimal for weakly communicating MDPs, in view of the minimax lower bound $\Omega(SAsp(^*)\epsilon^{-2})$. Existing work mainly focuses on ergodic MDPs and the results typically depend on $t_{mix},$ the worst-case mixing time induced by a policy. We remark that the diameter $D$ and mixing time $t_{mix}$ are both lower bounded by $sp(h^*)$, and $t_{mix}$ can be arbitrarily large for certain MDPs. On the technical side, our approach integrates two key ideas: learning an $\gamma$-discounted MDP as an approximation, and leveraging reference-advantage decomposition for variance in optimistic Q-learning. As recognized in prior work, a naive approximation by discounted MDPs results in suboptimal guarantees. A distinguishing feature of our method is maintaining estimates of value-difference between state pairs to provide a sharper bound on the variance of reference advantage. We also crucially use a careful choice of the discounted factor $\gamma$ to balance approximation error due to discounting and the statistical learning error, and we are able to maintain a good-quality reference value function with $O(SA)$ space complexity. 
    more » « less
  3. ; ; ; ; ; ; ( , Laser & Photonics Reviews)
    Abstract

    The incorporation of high‐performance optoelectronic devices into photonic neuromorphic processors can substantially accelerate computationally intensive matrix multiplication operations in machine learning (ML) algorithms. However, the conventional designs of individual devices and system are largely disconnected, and the system optimization is limited to the manual exploration of a small design space. Here, a device‐system end‐to‐end design methodology is reported to optimize a free‐space optical general matrix multiplication (GEMM) hardware accelerator by engineering a spatially reconfigurable array made from chalcogenide phase change materials. With a highly parallelized integrated hardware emulator with experimental information, the design of unit device to directly optimize GEMM calculation accuracy is achieved by exploring a large parameter space through reinforcement learning algorithms, including deep Q‐learning neural network, Bayesian optimization, and their cascaded approach. The algorithm‐generated physical quantities show a clear correlation between system performance metrics and device specifications. Furthermore, physics‐aware training approaches are employed to deploy optimized hardware to the tasks of image classification, materials discovery, and a closed‐loop design of optical ML accelerators. The demonstrated framework offers insights into the end‐to‐end and co‐design of optoelectronic devices and systems with reduced human supervision and domain knowledge barriers.

     
    more » « less
  4. ; ( , Proceedings of the 25th ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD), Anchorage, Alaska, August 2019)
    Machine learning models are bounded by the credibility of ground truth data used for both training and testing. Regardless of the problem domain, this ground truth annotation is objectively manual and tedious as it needs considerable amount of human intervention. With the advent of Active Learning with multiple annotators, the burden can be somewhat mitigated by actively acquiring labels of most informative data instances. However, multiple annotators with varying degrees of expertise poses new set of challenges in terms of quality of the label received and availability of the annotator. Due to limited amount of ground truth information addressing the variabilities of Activity of Daily Living (ADLs), activity recognition models using wearable and mobile devices are still not robust enough for real-world deployment. In this paper, we propose an active learning combined deep model which updates its network parameters based on the optimization of a joint loss function. We then propose a novel annotator selection model by exploiting the relationships among the users while considering their heterogeneity with respect to their expertise, physical and spatial context. Our proposed model leverages model-free deep reinforcement learning in a partially observable environment setting to capture the actionreward interaction among multiple annotators. Our experiments in real-world settings exhibit that our active deep model converges to optimal accuracy with fewer labeled instances and achieves 8% improvement in accuracy in fewer iterations. 
    more » « less
  5. ; ; ; ( , Proceedings of Machine Learning Research)
    Chaudhuri, Kamalika and (Ed.)
    We study the problem of reinforcement learning (RL) with low (policy) switching cost {—} a problem well-motivated by real-life RL applications in which deployments of new policies are costly and the number of policy updates must be low. In this paper, we propose a new algorithm based on stage-wise exploration and adaptive policy elimination that achieves a regret of $\widetilde{O}(\sqrt{H^4S^2AT})$ while requiring a switching cost of $O(HSA \log\log T)$. This is an exponential improvement over the best-known switching cost $O(H^2SA\log T)$ among existing methods with $\widetilde{O}(\mathrm{poly}(H,S,A)\sqrt{T})$ regret. In the above, $S,A$ denotes the number of states and actions in an $H$-horizon episodic Markov Decision Process model with unknown transitions, and $T$ is the number of steps. As a byproduct of our new techniques, we also derive a reward-free exploration algorithm with a switching cost of $O(HSA)$. Furthermore, we prove a pair of information-theoretical lower bounds which say that (1) Any no-regret algorithm must have a switching cost of $\Omega(HSA)$; (2) Any $\widetilde{O}(\sqrt{T})$ regret algorithm must incur a switching cost of $\Omega(HSA\log\log T)$. Both our algorithms are thus optimal in their switching costs. 
    more » « less
  • Citation Formats
  • MLA
  • APA
  • Chicago
  • BibTeX
  • Save / Share this Record
  • Send to Email