Deep reinforcement learning (DRL) remains challenging in tasks with sparse rewards. These sparse rewards often only indicate whether the task is partially or fully completed, meaning that many exploration actions must be performed before the agent obtains useful feedback. Hence, most existing DRL algorithms fail to learn feasible policies within a reasonable time frame. To overcome this problem, we develop an approach that exploits offline demonstration trajectories for faster and more efficient online RL in sparse reward settings. Our key insight is that by regarding offline demonstration trajectories as guidance, instead of imitating them, our method learns a policy whose state-action visitation marginal distribution matches that of offline demonstrations. Specifically, we introduce a novel trajectory distance based on maximum mean discrepancy (MMD) and formulate policy optimization as a distance-constrained optimization problem. Then, we show that this distance-constrained optimization problem can be reduced into a policy-gradient algorithm with shaped rewards learned from offline demonstrations. The proposed algorithm is evaluated on extensive discrete and continuous control tasks with sparse and deceptive rewards. The experimental results indicate that our proposed algorithm is significantly better than the baseline methods regarding diverse exploration and learning the optimal policy.
相關內容
We consider offline imitation learning (IL), which aims to mimic the expert's behavior from its demonstration without further interaction with the environment. One of the main challenges in offline IL is dealing with the limited support of expert demonstrations that cover only a small fraction of the state-action spaces. In this work, we consider offline IL, where expert demonstrations are limited but complemented by a larger set of sub-optimal demonstrations of lower expertise levels. Most of the existing offline IL methods developed for this setting are based on behavior cloning or distribution matching, where the aim is to match the occupancy distribution of the imitation policy with that of the expert policy. Such an approach often suffers from over-fitting, as expert demonstrations are limited to accurately represent any occupancy distribution. On the other hand, since sub-optimal sets are much larger, there is a high chance that the imitation policy is trained towards sub-optimal policies. In this paper, to address these issues, we propose a new approach based on inverse soft-Q learning, where a regularization term is added to the training objective, with the aim of aligning the learned rewards with a pre-assigned reward function that allocates higher weights to state-action pairs from expert demonstrations, and lower weights to those from lower expertise levels. On standard benchmarks, our inverse soft-Q learning significantly outperforms other offline IL baselines by a large margin.
Machine learning models have achieved great success in supervised learning tasks for end-to-end training, which requires a large amount of labeled data that is not always feasible. Recently, many practitioners have shifted to self-supervised learning methods that utilize cheap unlabeled data to learn a general feature extractor via pre-training, which can be further applied to personalized downstream tasks by simply training an additional linear layer with limited labeled data. However, such a process may also raise concerns regarding data poisoning attacks. For instance, indiscriminate data poisoning attacks, which aim to decrease model utility by injecting a small number of poisoned data into the training set, pose a security risk to machine learning models, but have only been studied for end-to-end supervised learning. In this paper, we extend the exploration of the threat of indiscriminate attacks on downstream tasks that apply pre-trained feature extractors. Specifically, we propose two types of attacks: (1) the input space attacks, where we modify existing attacks to directly craft poisoned data in the input space. However, due to the difficulty of optimization under constraints, we further propose (2) the feature targeted attacks, where we mitigate the challenge with three stages, firstly acquiring target parameters for the linear head; secondly finding poisoned features by treating the learned feature representations as a dataset; and thirdly inverting the poisoned features back to the input space. Our experiments examine such attacks in popular downstream tasks of fine-tuning on the same dataset and transfer learning that considers domain adaptation. Empirical results reveal that transfer learning is more vulnerable to our attacks. Additionally, input space attacks are a strong threat if no countermeasures are posed, but are otherwise weaker than feature targeted attacks.
Curriculum reinforcement learning (CRL) aims to speed up learning by gradually increasing the difficulty of a task, usually quantified by the achievable expected return. Inspired by the success of CRL in single-agent settings, a few works have attempted to apply CRL to multi-agent reinforcement learning (MARL) using the number of agents to control task difficulty. However, existing works typically use manually defined curricula such as a linear scheme. In this paper, we first apply state-of-the-art single-agent self-paced CRL to sparse reward MARL. Although with satisfying performance, we identify two potential flaws of the curriculum generated by existing reward-based CRL methods: (1) tasks with high returns may not provide informative learning signals and (2) the exacerbated credit assignment difficulty in tasks where more agents yield higher returns. Thereby, we further propose self-paced MARL (SPMARL) to prioritize tasks based on \textit{learning progress} instead of the episode return. Our method not only outperforms baselines in three challenging sparse-reward benchmarks but also converges faster than self-paced CRL.
This paper proposes SplitSGD, a new dynamic learning rate schedule for stochastic optimization. This method decreases the learning rate for better adaptation to the local geometry of the objective function whenever a stationary phase is detected, that is, the iterates are likely to bounce at around a vicinity of a local minimum. The detection is performed by splitting the single thread into two and using the inner product of the gradients from the two threads as a measure of stationarity. Owing to this simple yet provably valid stationarity detection, SplitSGD is easy-to-implement and essentially does not incur additional computational cost than standard SGD. Through a series of extensive experiments, we show that this method is appropriate for both convex problems and training (non-convex) neural networks, with performance compared favorably to other stochastic optimization methods. Importantly, this method is observed to be very robust with a set of default parameters for a wide range of problems and, moreover, can yield better generalization performance than other adaptive gradient methods such as Adam.
Data poisoning attacks manipulate training data to introduce unexpected behaviors into machine learning models at training time. For text-to-image generative models with massive training datasets, current understanding of poisoning attacks suggests that a successful attack would require injecting millions of poison samples into their training pipeline. In this paper, we show that poisoning attacks can be successful on generative models. We observe that training data per concept can be quite limited in these models, making them vulnerable to prompt-specific poisoning attacks, which target a model's ability to respond to individual prompts. We introduce Nightshade, an optimized prompt-specific poisoning attack where poison samples look visually identical to benign images with matching text prompts. Nightshade poison samples are also optimized for potency and can corrupt an Stable Diffusion SDXL prompt in <100 poison samples. Nightshade poison effects "bleed through" to related concepts, and multiple attacks can composed together in a single prompt. Surprisingly, we show that a moderate number of Nightshade attacks can destabilize general features in a text-to-image generative model, effectively disabling its ability to generate meaningful images. Finally, we propose the use of Nightshade` and similar tools as a last defense for content creators against web scrapers that ignore opt-out/do-not-crawl directives, and discuss possible implications for model trainers and content creators.
Despite the success of reinforcement learning from human feedback (RLHF) in aligning language models with human values, reward hacking, also termed reward overoptimization, remains a critical challenge, which primarily stems from limitations in reward modeling, i.e., generalizability of the reward model and inconsistency in the preference dataset. In this work, we tackle this problem from an information theoretic-perspective, and propose a generalizable and robust framework for reward modeling, namely InfoRM, by introducing a variational information bottleneck objective to filter out irrelevant information and developing a mechanism for model complexity modulation. Notably, we further identify a correlation between overoptimization and outliers in the latent space, establishing InfoRM as a promising tool for detecting reward overoptimization. Inspired by this finding, we propose the Integrated Cluster Deviation Score (ICDS), which quantifies deviations in the latent space, as an indicator of reward overoptimization to facilitate the development of online mitigation strategies. Extensive experiments on a wide range of settings and model scales (70M, 440M, 1.4B, and 7B) support the effectiveness of InfoRM. Further analyses reveal that InfoRM's overoptimization detection mechanism is effective, potentially signifying a notable advancement in the field of RLHF. Code will be released upon acceptance.
Despite the recent progress in deep learning, most approaches still go for a silo-like solution, focusing on learning each task in isolation: training a separate neural network for each individual task. Many real-world problems, however, call for a multi-modal approach and, therefore, for multi-tasking models. Multi-task learning (MTL) aims to leverage useful information across tasks to improve the generalization capability of a model. This thesis is concerned with multi-task learning in the context of computer vision. First, we review existing approaches for MTL. Next, we propose several methods that tackle important aspects of multi-task learning. The proposed methods are evaluated on various benchmarks. The results show several advances in the state-of-the-art of multi-task learning. Finally, we discuss several possibilities for future work.
Meta reinforcement learning (meta-RL) extracts knowledge from previous tasks and achieves fast adaptation to new tasks. Despite recent progress, efficient exploration in meta-RL remains a key challenge in sparse-reward tasks, as it requires quickly finding informative task-relevant experiences in both meta-training and adaptation. To address this challenge, we explicitly model an exploration policy learning problem for meta-RL, which is separated from exploitation policy learning, and introduce a novel empowerment-driven exploration objective, which aims to maximize information gain for task identification. We derive a corresponding intrinsic reward and develop a new off-policy meta-RL framework, which efficiently learns separate context-aware exploration and exploitation policies by sharing the knowledge of task inference. Experimental evaluation shows that our meta-RL method significantly outperforms state-of-the-art baselines on various sparse-reward MuJoCo locomotion tasks and more complex sparse-reward Meta-World tasks.
Graph Convolutional Networks (GCNs) have received increasing attention in recent machine learning. How to effectively leverage the rich structural information in complex graphs, such as knowledge graphs with heterogeneous types of entities and relations, is a primary open challenge in the field. Most GCN methods are either restricted to graphs with a homogeneous type of edges (e.g., citation links only), or focusing on representation learning for nodes only instead of jointly optimizing the embeddings of both nodes and edges for target-driven objectives. This paper addresses these limitations by proposing a novel framework, namely the GEneralized Multi-relational Graph Convolutional Networks (GEM-GCN), which combines the power of GCNs in graph-based belief propagation and the strengths of advanced knowledge-base embedding methods, and goes beyond. Our theoretical analysis shows that GEM-GCN offers an elegant unification of several well-known GCN methods as specific cases, with a new perspective of graph convolution. Experimental results on benchmark datasets show the advantageous performance of GEM-GCN over strong baseline methods in the tasks of knowledge graph alignment and entity classification.
Learning with limited data is a key challenge for visual recognition. Few-shot learning methods address this challenge by learning an instance embedding function from seen classes and apply the function to instances from unseen classes with limited labels. This style of transfer learning is task-agnostic: the embedding function is not learned optimally discriminative with respect to the unseen classes, where discerning among them is the target task. In this paper, we propose a novel approach to adapt the embedding model to the target classification task, yielding embeddings that are task-specific and are discriminative. To this end, we employ a type of self-attention mechanism called Transformer to transform the embeddings from task-agnostic to task-specific by focusing on relating instances from the test instances to the training instances in both seen and unseen classes. Our approach also extends to both transductive and generalized few-shot classification, two important settings that have essential use cases. We verify the effectiveness of our model on two standard benchmark few-shot classification datasets --- MiniImageNet and CUB, where our approach demonstrates state-of-the-art empirical performance.
小貼士
登錄享
相關主題
注冊地址： 北京市海淀區羊坊店路18號2幢3層301-191