A key component of model-based reinforcement learning (RL) is a dynamics model that predicts the outcomes of actions. Errors in this predictive model can degrade the performance of model-based controllers, and complex Markov decision processes (MDPs) can present exceptionally difficult prediction problems. To mitigate this issue, we propose predictable MDP abstraction (PMA): instead of training a predictive model on the original MDP, we train a model on a transformed MDP with a learned action space that only permits predictable, easy-to-model actions, while covering the original state-action space as much as possible. As a result, model learning becomes easier and more accurate, which allows robust, stable model-based planning or model-based RL. This transformation is learned in an unsupervised manner, before any task is specified by the user. Downstream tasks can then be solved with model-based control in a zero-shot fashion, without additional environment interactions. We theoretically analyze PMA and empirically demonstrate that PMA leads to significant improvements over prior unsupervised model-based RL approaches in a range of benchmark environments. Our code and videos are available at https://seohong.me/projects/pma/

Recent approaches have utilized self-supervised auxiliary tasks as representation learning to improve the performance and sample efficiency of vision-based reinforcement learning algorithms in single-agent settings. However, in multi-agent reinforcement learning (MARL), these techniques face challenges because each agent only receives partial observation from an environment influenced by others, resulting in correlated observations in the agent dimension. So it is necessary to consider agent-level information in representation learning for MARL. In this paper, we propose an effective framework called \textbf{M}ulti-\textbf{A}gent \textbf{M}asked \textbf{A}ttentive \textbf{C}ontrastive \textbf{L}earning (MA2CL), which encourages learning representation to be both temporal and agent-level predictive by reconstructing the masked agent observation in latent space. Specifically, we use an attention reconstruction model for recovering and the model is trained via contrastive learning. MA2CL allows better utilization of contextual information at the agent level, facilitating the training of MARL agents for cooperation tasks. Extensive experiments demonstrate that our method significantly improves the performance and sample efficiency of different MARL algorithms and outperforms other methods in various vision-based and state-based scenarios. Our code can be found in \url{https://github.com/ustchlsong/MA2CL}

How to train a generalizable meta-policy by continually learning a sequence of tasks? It is a natural human skill yet challenging to achieve by current reinforcement learning: the agent is expected to quickly adapt to new tasks (plasticity) meanwhile retaining the common knowledge from previous tasks (stability). We address it by "Continual Task Allocation via Sparse Prompting (CoTASP)", which learns over-complete dictionaries to produce sparse masks as prompts extracting a sub-network for each task from a meta-policy network. CoTASP trains a policy for each task by optimizing the prompts and the sub-network weights alternatively. The dictionary is then updated to align the optimized prompts with tasks' embedding, thereby capturing tasks' semantic correlations. Hence, relevant tasks share more neurons in the meta-policy network due to similar prompts while cross-task interference causing forgetting is effectively restrained. Given a meta-policy and dictionaries trained on previous tasks, new task adaptation reduces to highly efficient sparse prompting and sub-network finetuning. In experiments, CoTASP achieves a promising plasticity-stability trade-off without storing or replaying any past tasks' experiences. It outperforms existing continual and multi-task RL methods on all seen tasks, forgetting reduction, and generalization to unseen tasks.

Generative flow networks (GFlowNets) are a family of algorithms for training a sequential sampler of discrete objects under an unnormalized target density and have been successfully used for various probabilistic modeling tasks. Existing training objectives for GFlowNets are either local to states or transitions, or propagate a reward signal over an entire sampling trajectory. We argue that these alternatives represent opposite ends of a gradient bias-variance tradeoff and propose a way to exploit this tradeoff to mitigate its harmful effects. Inspired by the TD($\lambda$) algorithm in reinforcement learning, we introduce subtrajectory balance or SubTB($\lambda$), a GFlowNet training objective that can learn from partial action subsequences of varying lengths. We show that SubTB($\lambda$) accelerates sampler convergence in previously studied and new environments and enables training GFlowNets in environments with longer action sequences and sparser reward landscapes than what was possible before. We also perform a comparative analysis of stochastic gradient dynamics, shedding light on the bias-variance tradeoff in GFlowNet training and the advantages of subtrajectory balance.

Recommending novel content, which expands user horizons by introducing them to new interests, has been shown to improve users' long-term experience on recommendation platforms \cite{chen2021values}. Users however are not constantly looking to explore novel content. It is therefore crucial to understand their novelty-seeking intent and adjust the recommendation policy accordingly. Most existing literature models a user's propensity to choose novel content or to prefer a more diverse set of recommendations at individual interactions. Hierarchical structure, on the other hand, exists in a user's novelty-seeking intent, which is manifested as a static and intrinsic user preference for seeking novelty along with a dynamic session-based propensity. To this end, we propose a novel hierarchical reinforcement learning-based method to model the hierarchical user novelty-seeking intent, and to adapt the recommendation policy accordingly based on the extracted user novelty-seeking propensity. We further incorporate diversity and novelty-related measurement in the reward function of the hierarchical RL (HRL) agent to encourage user exploration \cite{chen2021values}. We demonstrate the benefits of explicitly modeling hierarchical user novelty-seeking intent in recommendations through extensive experiments on simulated and real-world datasets. In particular, we demonstrate that the effectiveness of our proposed hierarchical RL-based method lies in its ability to capture such hierarchically-structured intent. As a result, the proposed HRL model achieves superior performance on several public datasets, compared with state-of-art baselines.

Combining inputs from multiple sensor modalities effectively in reinforcement learning (RL) is an open problem. While many self-supervised representation learning approaches exist to improve performance and sample complexity for image-based RL, they usually neglect other available information, such as robot proprioception. However, using this proprioception for representation learning can help algorithms to focus on relevant aspects and guide them toward finding better representations. In this work, we systematically analyze representation learning for RL from multiple sensors by building on Recurrent State Space Models. We propose a combination of reconstruction-based and contrastive losses, which allows us to choose the most appropriate method for each sensor modality. We demonstrate the benefits of joint representations, particularly with distinct loss functions for each modality, for model-free and model-based RL on complex tasks. Those include tasks where the images contain distractions or occlusions and a new locomotion suite. We show that combining reconstruction-based and contrastive losses for joint representation learning improves performance significantly compared to a post hoc combination of image representations and proprioception and can also improve the quality of learned models for model-based RL.

While deep reinforcement learning (RL) algorithms have been successfully applied to many tasks, their inability to extrapolate and strong reliance on episodic resets inhibits their applicability to many real-world settings. For instance, in stochastic queueing problems, the state space can be unbounded and the agent may have to learn online without the system ever being reset to states the agent has seen before. In such settings, we show that deep RL agents can diverge into unseen states from which they can never recover due to the lack of resets, especially in highly stochastic environments. Towards overcoming this divergence, we introduce a Lyapunov-inspired reward shaping approach that encourages the agent to first learn to be stable (i.e. to achieve bounded cost) and then to learn to be optimal. We theoretically show that our reward shaping technique reduces the rate of divergence of the agent and empirically find that it prevents it. We further combine our reward shaping approach with a weight annealing scheme that gradually introduces optimality and log-transform of state inputs, and find that these techniques enable deep RL algorithms to learn high performing policies when learning online in unbounded state space domains.

We consider the reinforcement learning (RL) problem with general utilities which consists in maximizing a function of the state-action occupancy measure. Beyond the standard cumulative reward RL setting, this problem includes as particular cases constrained RL, pure exploration and learning from demonstrations among others. For this problem, we propose a simpler single-loop parameter-free normalized policy gradient algorithm. Implementing a recursive momentum variance reduction mechanism, our algorithm achieves $\tilde{\mathcal{O}}(\epsilon^{-3})$ and $\tilde{\mathcal{O}}(\epsilon^{-2})$ sample complexities for $\epsilon$-first-order stationarity and $\epsilon$-global optimality respectively, under adequate assumptions. We further address the setting of large finite state action spaces via linear function approximation of the occupancy measure and show a $\tilde{\mathcal{O}}(\epsilon^{-4})$ sample complexity for a simple policy gradient method with a linear regression subroutine.

In this work, we investigate the potential of improving multi-task training and also leveraging it for transferring in the reinforcement learning setting. We identify several challenges towards this goal and propose a transferring approach with a parameter-compositional formulation. We investigate ways to improve the training of multi-task reinforcement learning which serves as the foundation for transferring. Then we conduct a number of transferring experiments on various manipulation tasks. Experimental results demonstrate that the proposed approach can have improved performance in the multi-task training stage, and further show effective transferring in terms of both sample efficiency and performance.

This paper presents a comparison between two well-known deep Reinforcement Learning (RL) algorithms: Deep Q-Learning (DQN) and Proximal Policy Optimization (PPO) in a simulated production system. We utilize a Petri Net (PN)-based simulation environment, which was previously proposed in related work. The performance of the two algorithms is compared based on several evaluation metrics, including average percentage of correctly assembled and sorted products, average episode length, and percentage of successful episodes. The results show that PPO outperforms DQN in terms of all evaluation metrics. The study highlights the advantages of policy-based algorithms in problems with high-dimensional state and action spaces. The study contributes to the field of deep RL in context of production systems by providing insights into the effectiveness of different algorithms and their suitability for different tasks.