Most offline reinforcement learning (RL) methods suffer from the trade-off between improving the policy to surpass the behavior policy and constraining the policy to limit the deviation from the behavior policy as computing $Q$-values using out-of-distribution (OOD) actions will suffer from errors due to distributional shift. The recently proposed \textit{In-sample Learning} paradigm (i.e., IQL), which improves the policy by quantile regression using only data samples, shows great promise because it learns an optimal policy without querying the value function of any unseen actions. However, it remains unclear how this type of method handles the distributional shift in learning the value function. In this work, we make a key finding that the in-sample learning paradigm arises under the \textit{Implicit Value Regularization} (IVR) framework. This gives a deeper understanding of why the in-sample learning paradigm works, i.e., it applies implicit value regularization to the policy. Based on the IVR framework, we further propose two practical algorithms, Sparse $Q$-learning (SQL) and Exponential $Q$-learning (EQL), which adopt the same value regularization used in existing works, but in a complete in-sample manner. Compared with IQL, we find that our algorithms introduce sparsity in learning the value function, making them more robust in noisy data regimes. We also verify the effectiveness of SQL and EQL on D4RL benchmark datasets and show the benefits of in-sample learning by comparing them with CQL in small data regimes.

Applying reinforcement learning (RL) to sparse reward domains is notoriously challenging due to insufficient guiding signals. Common RL techniques for addressing such domains include (1) learning from demonstrations and (2) curriculum learning. While these two approaches have been studied in detail, they have rarely been considered together. This paper aims to do so by introducing a principled task phasing approach that uses demonstrations to automatically generate a curriculum sequence. Using inverse RL from (suboptimal) demonstrations we define a simple initial task. Our task phasing approach then provides a framework to gradually increase the complexity of the task all the way to the target task, while retuning the RL agent in each phasing iteration. Two approaches for phasing are considered: (1) gradually increasing the proportion of time steps an RL agent is in control, and (2) phasing out a guiding informative reward function. We present conditions that guarantee the convergence of these approaches to an optimal policy. Experimental results on 3 sparse reward domains demonstrate that our task phasing approaches outperform state-of-the-art approaches with respect to asymptotic performance.

Learning various motor skills for quadrupedal robots is a challenging problem that requires careful design of task-specific mathematical models or reward descriptions. In this work, we propose to learn a single capable policy using deep reinforcement learning by imitating a large number of reference motions, including walking, turning, pacing, jumping, sitting, and lying. On top of the existing motion imitation framework, we first carefully design the observation space, the action space, and the reward function to improve the scalability of the learning as well as the robustness of the final policy. In addition, we adopt a novel adaptive motion sampling (AMS) method, which maintains a balance between successful and unsuccessful behaviors. This technique allows the learning algorithm to focus on challenging motor skills and avoid catastrophic forgetting. We demonstrate that the learned policy can exhibit diverse behaviors in simulation by successfully tracking both the training dataset and out-of-distribution trajectories. We also validate the importance of the proposed learning formulation and the adaptive motion sampling scheme by conducting experiments.

Knowledge distillation (KD) has demonstrated its effectiveness to boost the performance of graph neural networks (GNNs), where its goal is to distill knowledge from a deeper teacher GNN into a shallower student GNN. However, it is actually difficult to train a satisfactory teacher GNN due to the well-known over-parametrized and over-smoothing issues, leading to invalid knowledge transfer in practical applications. In this paper, we propose the first Free-direction Knowledge Distillation framework via Reinforcement learning for GNNs, called FreeKD, which is no longer required to provide a deeper well-optimized teacher GNN. The core idea of our work is to collaboratively build two shallower GNNs in an effort to exchange knowledge between them via reinforcement learning in a hierarchical way. As we observe that one typical GNN model often has better and worse performances at different nodes during training, we devise a dynamic and free-direction knowledge transfer strategy that consists of two levels of actions: 1) node-level action determines the directions of knowledge transfer between the corresponding nodes of two networks; and then 2) structure-level action determines which of the local structures generated by the node-level actions to be propagated. In essence, our FreeKD is a general and principled framework which can be naturally compatible with GNNs of different architectures. Extensive experiments on five benchmark datasets demonstrate our FreeKD outperforms two base GNNs in a large margin, and shows its efficacy to various GNNs. More surprisingly, our FreeKD has comparable or even better performance than traditional KD algorithms that distill knowledge from a deeper and stronger teacher GNN.

It is important for deep reinforcement learning (DRL) algorithms to transfer their learned policies to new environments that have different visual inputs. In this paper, we introduce Prompt based Proximal Policy Optimization ($P^{3}O$), a three-stage DRL algorithm that transfers visual representations from a target to a source environment by applying prompting. The process of $P^{3}O$ consists of three stages: pre-training, prompting, and predicting. In particular, we specify a prompt-transformer for representation conversion and propose a two-step training process to train the prompt-transformer for the target environment, while the rest of the DRL pipeline remains unchanged. We implement $P^{3}O$ and evaluate it on the OpenAI CarRacing video game. The experimental results show that $P^{3}O$ outperforms the state-of-the-art visual transferring schemes. In particular, $P^{3}O$ allows the learned policies to perform well in environments with different visual inputs, which is much more effective than retraining the policies in these environments.

Although reinforcement learning has seen tremendous success recently, this kind of trial-and-error learning can be impractical or inefficient in complex environments. The use of demonstrations, on the other hand, enables agents to benefit from expert knowledge rather than having to discover the best action to take through exploration. In this survey, we discuss the advantages of using demonstrations in sequential decision making, various ways to apply demonstrations in learning-based decision making paradigms (for example, reinforcement learning and planning in the learned models), and how to collect the demonstrations in various scenarios. Additionally, we exemplify a practical pipeline for generating and utilizing demonstrations in the recently proposed ManiSkill robot learning benchmark.

In this work, we aim to learn dexterous manipulation of deformable objects using multi-fingered hands. Reinforcement learning approaches for dexterous rigid object manipulation would struggle in this setting due to the complexity of physics interaction with deformable objects. At the same time, previous trajectory optimization approaches with differentiable physics for deformable manipulation would suffer from local optima caused by the explosion of contact modes from hand-object interactions. To address these challenges, we propose DexDeform, a principled framework that abstracts dexterous manipulation skills from human demonstration and refines the learned skills with differentiable physics. Concretely, we first collect a small set of human demonstrations using teleoperation. And we then train a skill model using demonstrations for planning over action abstractions in imagination. To explore the goal space, we further apply augmentations to the existing deformable shapes in demonstrations and use a gradient optimizer to refine the actions planned by the skill model. Finally, we adopt the refined trajectories as new demonstrations for finetuning the skill model. To evaluate the effectiveness of our approach, we introduce a suite of six challenging dexterous deformable object manipulation tasks. Compared with baselines, DexDeform is able to better explore and generalize across novel goals unseen in the initial human demonstrations.

We provide open, transparent implementation and assessment of Google Brain's deep reinforcement learning approach to macro placement and its Circuit Training (CT) implementation in GitHub. We implement in open source key "blackbox" elements of CT, and clarify discrepancies between CT and Nature paper. New testcases on open enablements are developed and released. We assess CT alongside multiple alternative macro placers, with all evaluation flows and related scripts public in GitHub. Our experiments also encompass academic mixed-size placement benchmarks, as well as ablation and stability studies. We comment on the impact of Nature and CT, as well as directions for future research.

Adversarial robustness assessment for video recognition models has raised concerns owing to their wide applications on safety-critical tasks. Compared with images, videos have much high dimension, which brings huge computational costs when generating adversarial videos. This is especially serious for the query-based black-box attacks where gradient estimation for the threat models is usually utilized, and high dimensions will lead to a large number of queries. To mitigate this issue, we propose to simultaneously eliminate the temporal and spatial redundancy within the video to achieve an effective and efficient gradient estimation on the reduced searching space, and thus query number could decrease. To implement this idea, we design the novel Adversarial spatial-temporal Focus (AstFocus) attack on videos, which performs attacks on the simultaneously focused key frames and key regions from the inter-frames and intra-frames in the video. AstFocus attack is based on the cooperative Multi-Agent Reinforcement Learning (MARL) framework. One agent is responsible for selecting key frames, and another agent is responsible for selecting key regions. These two agents are jointly trained by the common rewards received from the black-box threat models to perform a cooperative prediction. By continuously querying, the reduced searching space composed of key frames and key regions is becoming precise, and the whole query number becomes less than that on the original video. Extensive experiments on four mainstream video recognition models and three widely used action recognition datasets demonstrate that the proposed AstFocus attack outperforms the SOTA methods, which is prevenient in fooling rate, query number, time, and perturbation magnitude at the same.

Most successes in robotic manipulation have been restricted to single-arm gripper robots, whose low dexterity limits the range of solvable tasks to pick-and-place, inser-tion, and object rearrangement. More complex tasks such as assembly require dual and multi-arm platforms, but entail a suite of unique challenges such as bi-arm coordination and collision avoidance, robust grasping, and long-horizon planning. In this work we investigate the feasibility of training deep reinforcement learning (RL) policies in simulation and transferring them to the real world (Sim2Real) as a generic methodology for obtaining performant controllers for real-world bi-manual robotic manipulation tasks. As a testbed for bi-manual manipulation, we develop the U-Shape Magnetic BlockAssembly Task, wherein two robots with parallel grippers must connect 3 magnetic blocks to form a U-shape. Without manually-designed controller nor human demonstrations, we demonstrate that with careful Sim2Real considerations, our policies trained with RL in simulation enable two xArm6 robots to solve the U-shape assembly task with a success rate of above90% in simulation, and 50% on real hardware without any additional real-world fine-tuning. Through careful ablations,we highlight how each component of the system is critical for such simple and successful policy learning and transfer,including task specification, learning algorithm, direct joint-space control, behavior constraints, perception and actuation noises, action delays and action interpolation. Our results present a significant step forward for bi-arm capability on real hardware, and we hope our system can inspire future research on deep RL and Sim2Real transfer of bi-manualpolicies, drastically scaling up the capability of real-world robot manipulators.