First-person object-interaction tasks in high-fidelity, 3D, simulated environments such as the AI2Thor virtual home-environment pose significant sample-efficiency challenges for reinforcement learning (RL) agents learning from sparse task rewards. To alleviate these challenges, prior work has provided extensive supervision via a combination of reward-shaping, ground-truth object-information, and expert demonstrations. In this work, we show that one can learn object-interaction tasks from scratch without supervision by learning an attentive object-model as an auxiliary task during task learning with an object-centric relational RL agent. Our key insight is that learning an object-model that incorporates object-attention into forward prediction provides a dense learning signal for unsupervised representation learning of both objects and their relationships. This, in turn, enables faster policy learning for an object-centric relational RL agent. We demonstrate our agent by introducing a set of challenging object-interaction tasks in the AI2Thor environment where learning with our attentive object-model is key to strong performance. Specifically, we compare our agent and relational RL agents with alternative auxiliary tasks to a relational RL agent equipped with ground-truth object-information, and show that learning with our object-model best closes the performance gap in terms of both learning speed and maximum success rate. Additionally, we find that incorporating object-attention into an object-model's forward predictions is key to learning representations which capture object-category and object-state.
Model-based reinforcement learning (RL) has shown great potential in various control tasks in terms of both sample-efficiency and final performance. However, learning a generalizable dynamics model robust to changes in dynamics remains a challenge since the target transition dynamics follow a multi-modal distribution. In this paper, we present a new model-based RL algorithm, coined trajectory-wise multiple choice learning, that learns a multi-headed dynamics model for dynamics generalization. The main idea is updating the most accurate prediction head to specialize each head in certain environments with similar dynamics, i.e., clustering environments. Moreover, we incorporate context learning, which encodes dynamics-specific information from past experiences into the context latent vector, enabling the model to perform online adaptation to unseen environments. Finally, to utilize the specialized prediction heads more effectively, we propose an adaptive planning method, which selects the most accurate prediction head over a recent experience. Our method exhibits superior zero-shot generalization performance across a variety of control tasks, compared to state-of-the-art RL methods. Source code and videos are available at https://sites.google.com/view/trajectory-mcl.
In an effort to overcome limitations of reward-driven feature learning in deep reinforcement learning (RL) from images, we propose decoupling representation learning from policy learning. To this end, we introduce a new unsupervised learning (UL) task, called Augmented Temporal Contrast (ATC), which trains a convolutional encoder to associate pairs of observations separated by a short time difference, under image augmentations and using a contrastive loss. In online RL experiments, we show that training the encoder exclusively using ATC matches or outperforms end-to-end RL in most environments. Additionally, we benchmark several leading UL algorithms by pre-training encoders on expert demonstrations and using them, with weights frozen, in RL agents; we find that agents using ATC-trained encoders outperform all others. We also train multi-task encoders on data from multiple environments and show generalization to different downstream RL tasks. Finally, we ablate components of ATC, and introduce a new data augmentation to enable replay of (compressed) latent images from pre-trained encoders when RL requires augmentation. Our experiments span visually diverse RL benchmarks in DeepMind Control, DeepMind Lab, and Atari, and our complete code is available at https://github.com/astooke/rlpyt/tree/master/rlpyt/ul.
Despite the significant progress of deep reinforcement learning (RL) in solving sequential decision making problems, RL agents often overfit to training environments and struggle to adapt to new, unseen environments. This prevents robust applications of RL in real world situations, where system dynamics may deviate wildly from the training settings. In this work, our primary contribution is to propose an information theoretic regularization objective and an annealing-based optimization method to achieve better generalization ability in RL agents. We demonstrate the extreme generalization benefits of our approach in different domains ranging from maze navigation to robotic tasks; for the first time, we show that agents can generalize to test parameters more than 10 standard deviations away from the training parameter distribution. This work provides a principled way to improve generalization in RL by gradually removing information that is redundant for task-solving; it opens doors for the systematic study of generalization from training to extremely different testing settings, focusing on the established connections between information theory and machine learning.
Experience replay, which enables the agents to remember and reuse experience from the past, plays a significant role in the success of off-policy reinforcement learning (RL). To utilize the experience replay efficiently, experience transitions should be sampled with consideration of their significance, such that the known prioritized experience replay (PER) further allows to sample more important experience. Yet, the conventional PER may result in generating highly biased samples due to considering a single metric such as TD-error and computing the sampling rate independently for each experience. To tackle this issue, we propose a Neural Experience Replay Sampler (NERS), which adaptively evaluates the relative importance of a sampled transition by obtaining context from not only its (local) values that characterize itself such as TD-error or the raw features but also other (global) transitions. We validate our framework on multiple benchmark tasks for both continuous and discrete controls and show that the proposed framework significantly improves the performance of various off-policy RL methods. Further analysis confirms that the improvements indeed come from the use of diverse features and the consideration of the relative importance of experiences.
Model-free deep reinforcement learning (RL) has been successful in a range of challenging domains. However, there are some remaining issues, such as stabilizing the optimization of nonlinear function approximators, preventing error propagation due to the Bellman backup in Q-learning, and efficient exploration. To mitigate these issues, we present SUNRISE, a simple unified ensemble method, which is compatible with various off-policy RL algorithms. SUNRISE integrates three key ingredients: (a) bootstrap with random initialization which improves the stability of the learning process by training a diverse ensemble of agents, (b) weighted Bellman backups, which prevent error propagation in Q-learning by reweighing sample transitions based on uncertainty estimates from the ensembles, and (c) an inference method that selects actions using highest upper-confidence bounds for efficient exploration. Our experiments show that SUNRISE significantly improves the performance of existing off-policy RL algorithms, such as Soft Actor-Critic and Rainbow DQN, for both continuous and discrete control tasks on both low-dimensional and high-dimensional environments. Our training code is available at https://github.com/pokaxpoka/sunrise.
Model-based reinforcement learning (RL) enjoys several benefits, such as data-efficiency and planning, by learning a model of the environment's dynamics. However, learning a global model that can generalize across different dynamics is a challenging task. To tackle this problem, we decompose the task of learning a global dynamics model into two stages: (a) learning a context latent vector that captures the local dynamics, then (b) predicting the next state conditioned on it. In order to encode dynamics-specific information into the context latent vector, we introduce a novel loss function that encourages the context latent vector to be useful for predicting both forward and backward dynamics. The proposed method achieves superior generalization ability across various simulated robotics and control tasks, compared to existing RL schemes.
Learning from visual observations is a fundamental yet challenging problem in reinforcement learning (RL). Although algorithmic advancements combined with convolutional neural networks have proved to be a recipe for success, current methods are still lacking on two fronts: (a) sample efficiency of learning and (b) generalization to new environments. To this end, we present RAD: Reinforcement Learning with Augmented Data, a simple plug-and-play module that can enhance any RL algorithm. We show that data augmentations such as random crop, color jitter, patch cutout, and random convolutions can enable simple RL algorithms to match and even outperform complex state-of-the-art methods across common benchmarks in terms of data-efficiency, generalization, and wall-clock speed. We find that data diversity alone can make agents focus on meaningful information from high-dimensional observations without any changes to the reinforcement learning method. On the DeepMind Control Suite, we show that RAD is state-of-the-art in terms of data-efficiency and performance across 15 environments. We further demonstrate that RAD can significantly improve the test-time generalization on several OpenAI ProcGen benchmarks. Finally, our customized data augmentation modules enable faster wall-clock speed compared to competing RL techniques. Our RAD module and training code are available at https://www.github.com/MishaLaskin/rad.