Recently, deep reinforcement learning (DRL) has achieved promising results in solving online 3D Bin Packing Problems (3D-BPP). However, these DRL-based policies may perform poorly on new instances due to distribution shift. Besides generalization, we also consider adaptation, completely overlooked by previous work, which aims at rapidly finetuning these policies to a new test distribution. To tackle both generalization and adaptation issues, we propose Adaptive Selection After Pruning (ASAP), which decomposes a solver's decision-making into two policies, one for pruning and one for selection. The role of the pruning policy is to remove inherently bad actions, which allows the selection policy to choose among the remaining most valuable actions. To learn these policies, we propose a training scheme based on a meta-learning phase of both policies followed by a finetuning phase of the sole selection policy to rapidly adapt it to a test distribution. Our experiments demonstrate that ASAP exhibits excellent generalization and adaptation capabilities on in-distribution and out-of-distribution instances under both discrete and continuous setup.

A major challenge in Reinforcement Learning (RL) is the difficulty of learning an optimal policy from sparse rewards. Prior works enhance online RL with conventional Imitation Learning (IL) via a handcrafted auxiliary objective, at the cost of restricting the RL policy to be sub-optimal when the offline data is generated by a non-expert policy. Instead, to better leverage valuable information in offline data, we develop Generalized Imitation Learning from Demonstration (GILD), which meta-learns an objective that distills knowledge from offline data and instills intrinsic motivation towards the optimal policy. Distinct from prior works that are exclusive to a specific RL algorithm, GILD is a flexible module intended for diverse vanilla off-policy RL algorithms. In addition, GILD introduces no domain-specific hyperparameter and minimal increase in computational cost. In four challenging MuJoCo tasks with sparse rewards, we show that three RL algorithms enhanced with GILD significantly outperform state-of-the-art methods.

A major bottleneck in imitation learning is the requirement of a large number of expert demonstrations, which can be expensive or inaccessible. Learning from supplementary demonstrations without strict quality requirements has emerged as a powerful paradigm to address this challenge. However, previous methods often fail to fully utilize their potential by discarding non-expert data. Our key insight is that even demonstrations that fall outside the expert distribution but outperform the learned policy can enhance policy performance. To utilize this potential, we propose a novel approach named imitation learning via meta-learning an action ranker (ILMAR). ILMAR implements weighted behavior cloning (weighted BC) on a limited set of expert demonstrations along with supplementary demonstrations. It utilizes the functional of the advantage function to selectively integrate knowledge from the supplementary demonstrations. To make more effective use of supplementary demonstrations, we introduce meta-goal in ILMAR to optimize the functional of the advantage function by explicitly minimizing the distance between the current policy and the expert policy. Comprehensive experiments using extensive tasks demonstrate that ILMAR significantly outperforms previous methods in handling suboptimal demonstrations. Code is available at https://github.com/F-GOD6/ILMAR.

In practice, reinforcement learning (RL) agents are often trained with a possibly imperfect proxy reward function, which may lead to a human-agent alignment issue (i.e., the learned policy either converges to non-optimal performance with low cumulative rewards, or achieves high cumulative rewards but in undesired manner). To tackle this issue, we consider a framework where a human labeler can provide additional feedback in the form of corrective actions, which expresses the labeler's action preferences although this feedback may possibly be imperfect as well. In this setting, to obtain a better-aligned policy guided by both learning signals, we propose a novel value-based deep RL algorithm called Iterative learning from Corrective actions and Proxy rewards (ICoPro), which cycles through three phases: (1) Solicit sparse corrective actions from a human labeler on the agent's demonstrated trajectories; (2) Incorporate these corrective actions into the Q-function using a margin loss to enforce adherence to labeler's preferences; (3) Train the agent with standard RL losses regularized with a margin loss to learn from proxy rewards and propagate the Q-values learned from human feedback. Moreover, another novel design in our approach is to integrate pseudo-labels from the target Q-network to reduce human labor and further stabilize training. We experimentally validate our proposition on a variety of tasks (Atari games and autonomous driving on highway). On the one hand, using proxy rewards with different levels of imperfection, our method can better align with human preferences and is more sample-efficient than baseline methods. On the other hand, facing corrective actions with different types of imperfection, our method can overcome the non-optimality of this feedback thanks to the guidance from proxy reward.

Data augmentation is widely applied and has shown its benefits in different machine learning tasks. However, as recently observed in some downstream tasks, data augmentation may introduce an unfair impact on classifications. While it can improve the performance of some classes, it can actually be detrimental for other classes, which can be problematic in some application domains. In this paper, to counteract this phenomenon, we propose a FAir Classification approach with a Two-player game (FACT). We first formulate the training of a classifier with data augmentation as a fair optimization problem, which can be further written as an adversarial two-player game. Following this formulation, we propose a novel multiplicative weight optimization algorithm, for which we theoretically prove that it can converge to a solution that is fair over classes. Interestingly, our formulation also reveals that this fairness issue over classes is not due to data augmentation only, but is in fact a general phenomenon. Our empirical experiments demonstrate that the performance of our learned classifiers is indeed more fairly distributed over classes in five datasets, with only limited impact on the average accuracy.

Recently, deep reinforcement learning has shown promising results for learning fast heuristics to solve routing problems. Meanwhile, most of the solvers suffer from generalizing to an unseen distribution or distributions with different scales. To address this issue, we propose a novel architecture, called Invariant Nested View Transformer (INViT), which is designed to enforce a nested design together with invariant views inside the encoders to promote the generalizability of the learned solver. It applies a modified policy gradient algorithm enhanced with data augmentations. We demonstrate that the proposed INViT achieves a dominant generalization performance on both TSP and CVRP problems with various distributions and different problem scales.

To improve the sample efficiency of vision-based deep reinforcement learning (RL), we propose a novel method, called SPIRL, to automatically extract important patches from input images. Following Masked Auto-Encoders, SPIRL is based on Vision Transformer models pre-trained in a self-supervised fashion to reconstruct images from randomly-sampled patches. These pre-trained models can then be exploited to detect and select salient patches, defined as hard to reconstruct from neighboring patches. In RL, the SPIRL agent processes selected salient patches via an attention module. We empirically validate SPIRL on Atari games to test its data-efficiency against relevant state-of-the-art methods, including some traditional model-based methods and keypoint-based models. In addition, we analyze our model's interpretability capabilities.

Reinforcement learning from human feedback (RLHF) is a variant of reinforcement learning (RL) that learns from human feedback instead of relying on an engineered reward function. Building on prior work on the related setting of preference-based reinforcement learning (PbRL), it stands at the intersection of artificial intelligence and human-computer interaction. This positioning offers a promising avenue to enhance the performance and adaptability of intelligent systems while also improving the alignment of their objectives with human values. The training of Large Language Models (LLMs) has impressively demonstrated this potential in recent years, where RLHF played a decisive role in targeting the model's capabilities toward human objectives. This article provides a comprehensive overview of the fundamentals of RLHF, exploring the intricate dynamics between machine agents and human input. While recent focus has been on RLHF for LLMs, our survey adopts a broader perspective, examining the diverse applications and wide-ranging impact of the technique. We delve into the core principles that underpin RLHF, shedding light on the symbiotic relationship between algorithms and human feedback, and discuss the main research trends in the field. By synthesizing the current landscape of RLHF research, this article aims to provide researchers as well as practitioners with a comprehensive understanding of this rapidly growing field of research.

When applying reinforcement learning (RL) to a new problem, reward engineering is a necessary, but often difficult and error-prone task a system designer has to face. To avoid this step, we propose LR4GPM, a novel (deep) RL method that can optimize a global performance metric, which is supposed to be available as part of the problem description. LR4GPM alternates between two phases: (1) learning a (possibly vector) reward function used to fit the performance metric, and (2) training a policy to optimize an approximation of this performance metric based on the learned rewards. Such RL training is not straightforward since both the reward function and the policy are trained using non-stationary data. To overcome this issue, we propose several training tricks. We demonstrate the efficiency of LR4GPM on several domains. Notably, LR4GPM outperforms the winner of a recent autonomous driving competition organized at DAI'2020.

We propose an efficient interpretable neuro-symbolic model to solve Inductive Logic Programming (ILP) problems. In this model, which is built from a set of meta-rules organised in a hierarchical structure, first-order rules are invented by learning embeddings to match facts and body predicates of a meta-rule. To instantiate it, we specifically design an expressive set of generic meta-rules, and demonstrate they generate a consequent fragment of Horn clauses. During training, we inject a controlled \pw{Gumbel} noise to avoid local optima and employ interpretability-regularization term to further guide the convergence to interpretable rules. We empirically validate our model on various tasks (ILP, visual genome, reinforcement learning) against several state-of-the-art methods.