To generalize across tasks, an agent should acquire knowledge from past tasks that facilitate adaptation and exploration in future tasks. We focus on the problem of in-context adaptation and exploration, where an agent only relies on context, i.e., history of states, actions and/or rewards, rather than gradient-based updates. Posterior sampling (extension of Thompson sampling) is a promising approach, but it requires Bayesian inference and dynamic programming, which often involve unknowns (e.g., a prior) and costly computations. To address these difficulties, we use a transformer to learn an inference process from training tasks and consider a hypothesis space of partial models, represented as small Markov decision processes that are cheap for dynamic programming. In our version of the Symbolic Alchemy benchmark, our method's adaptation speed and exploration-exploitation balance approach those of an exact posterior sampling oracle. We also show that even though partial models exclude relevant information from the environment, they can nevertheless lead to good policies.
相關內容
Imitation learning from demonstrations (ILD) aims to alleviate numerous shortcomings of reinforcement learning through the use of demonstrations. However, in most real-world applications, expert action guidance is absent, making the use of ILD impossible. Instead, we consider imitation learning from observations (ILO), where no expert actions are provided, making it a significantly more challenging problem to address. Existing methods often employ on-policy learning, which is known to be sample-costly. This paper presents SEILO, a novel sample-efficient on-policy algorithm for ILO, that combines standard adversarial imitation learning with inverse dynamics modeling. This approach enables the agent to receive feedback from both the adversarial procedure and a behavior cloning loss. We empirically demonstrate that our proposed algorithm requires fewer interactions with the environment to achieve expert performance compared to other state-of-the-art on-policy ILO and ILD methods.
Pre-training & fine-tuning is a prevalent paradigm in computer vision (CV). Recently, parameter-efficient transfer learning (PETL) methods have shown promising performance in transferring knowledge from pre-trained models with only a few trainable parameters. Despite their success, the existing PETL methods in CV can be computationally expensive and require large amounts of memory and time cost during training, which limits low-resource users from conducting research and applications on large models. In this work, we propose Parameter, Memory, and Time Efficient Visual Adapter ($\mathrm{E^3VA}$) tuning to address this issue. We provide a gradient backpropagation highway for low-rank adapters which removes large gradient computations for the frozen pre-trained parameters, resulting in substantial savings of training memory and training time. Furthermore, we optimise the $\mathrm{E^3VA}$ structure for dense predictions tasks to promote model performance. Extensive experiments on COCO, ADE20K, and Pascal VOC benchmarks show that $\mathrm{E^3VA}$ can save up to 62.2% training memory and 26.2% training time on average, while achieving comparable performance to full fine-tuning and better performance than most PETL methods. Note that we can even train the Swin-Large-based Cascade Mask RCNN on GTX 1080Ti GPUs with less than 1.5% trainable parameters.
Learning to control unknown nonlinear dynamical systems is a fundamental problem in reinforcement learning and control theory. A commonly applied approach is to first explore the environment (exploration), learn an accurate model of it (system identification), and then compute an optimal controller with the minimum cost on this estimated system (policy optimization). While existing work has shown that it is possible to learn a uniformly good model of the system~\citep{mania2020active}, in practice, if we aim to learn a good controller with a low cost on the actual system, certain system parameters may be significantly more critical than others, and we therefore ought to focus our exploration on learning such parameters. In this work, we consider the setting of nonlinear dynamical systems and seek to formally quantify, in such settings, (a) which parameters are most relevant to learning a good controller, and (b) how we can best explore so as to minimize uncertainty in such parameters. Inspired by recent work in linear systems~\citep{wagenmaker2021task}, we show that minimizing the controller loss in nonlinear systems translates to estimating the system parameters in a particular, task-dependent metric. Motivated by this, we develop an algorithm able to efficiently explore the system to reduce uncertainty in this metric, and prove a lower bound showing that our approach learns a controller at a near-instance-optimal rate. Our algorithm relies on a general reduction from policy optimization to optimal experiment design in arbitrary systems, and may be of independent interest. We conclude with experiments demonstrating the effectiveness of our method in realistic nonlinear robotic systems.
Code comment generation aims at generating natural language descriptions for a code snippet to facilitate developers' program comprehension activities. Despite being studied for a long time, a bottleneck for existing approaches is that given a code snippet, they can only generate one comment while developers usually need to know information from diverse perspectives such as what is the functionality of this code snippet and how to use it. To tackle this limitation, this study empirically investigates the feasibility of utilizing large language models (LLMs) to generate comments that can fulfill developers' diverse intents. Our intuition is based on the facts that (1) the code and its pairwise comment are used during the pre-training process of LLMs to build the semantic connection between the natural language and programming language, and (2) comments in the real-world projects, which are collected for the pre-training, usually contain different developers' intents. We thus postulate that the LLMs can already understand the code from different perspectives after the pre-training. Indeed, experiments on two large-scale datasets demonstrate the rationale of our insights: by adopting the in-context learning paradigm and giving adequate prompts to the LLM (e.g., providing it with ten or more examples), the LLM can significantly outperform a state-of-the-art supervised learning approach on generating comments with multiple intents. Results also show that customized strategies for constructing the prompts and post-processing strategies for reranking the results can both boost the LLM's performances, which shed light on future research directions for using LLMs to achieve comment generation.
We study social learning dynamics where the agents collectively follow a simple multi-armed bandit protocol. Agents arrive sequentially, choose arms and receive associated rewards. Each agent observes the full history (arms and rewards) of the previous agents, and there are no private signals. While collectively the agents face exploration-exploitation tradeoff, each agent acts myopically, without regards to exploration. Motivating scenarios concern reviews and ratings on online platforms. We allow a wide range of myopic behaviors that are consistent with (parameterized) confidence intervals, including the "unbiased" behavior as well as various behaviorial biases. While extreme versions of these behaviors correspond to well-known bandit algorithms, we prove that more moderate versions lead to stark exploration failures, and consequently to regret rates that are linear in the number of agents. We provide matching upper bounds on regret by analyzing "moderately optimistic" agents. As a special case of independent interest, we obtain a general result on failure of the greedy algorithm in multi-armed bandits. This is the first such result in the literature, to the best of our knowledge.
Robots operating alongside humans often encounter unfamiliar environments that make autonomous task completion challenging. Though improving models and increasing dataset size can enhance a robot's performance in unseen environments, dataset generation and model refinement may be impractical in every unfamiliar environment. Approaches that utilize human demonstration through manual operation can aid in generalizing to these unfamiliar environments, but often require significant human effort and expertise to achieve satisfactory task performance. To address these challenges, we propose leveraging part-time human interaction for redirection of robots during failed task execution. We train a lightweight help policy that allows robots to learn when to proceed autonomously or request human assistance at times of uncertainty. By incorporating part-time human intervention, robots recover quickly from their mistakes. Our best performing policy yields a 20 percent increase in path-length weighted success with only a 21 percent human interaction ratio. This approach provides a practical means for robots to interact and learn from humans in real-world settings, facilitating effective task completion without the need for significant human intervention.
Large pre-trained speech models are widely used as the de-facto paradigm, especially in scenarios when there is a limited amount of labeled data available. However, finetuning all parameters from the self-supervised learned model can be computationally expensive, and becomes infeasiable as the size of the model and the number of downstream tasks scales. In this paper, we propose a novel approach called Two Parallel Adapter (TPA) that is inserted into the conformer-based model pre-trained model instead. TPA is based on systematic studies of the residual adapter, a popular approach for finetuning a subset of parameters. We evaluate TPA on various public benchmarks and experiment results demonstrates its superior performance, which is close to the full finetuning on different datasets and speech tasks. These results show that TPA is an effective and efficient approach for serving large pre-trained speech models. Ablation studies show that TPA can also be pruned, especially for lower blocks.
Unsupervised domain adaptation has recently emerged as an effective paradigm for generalizing deep neural networks to new target domains. However, there is still enormous potential to be tapped to reach the fully supervised performance. In this paper, we present a novel active learning strategy to assist knowledge transfer in the target domain, dubbed active domain adaptation. We start from an observation that energy-based models exhibit free energy biases when training (source) and test (target) data come from different distributions. Inspired by this inherent mechanism, we empirically reveal that a simple yet efficient energy-based sampling strategy sheds light on selecting the most valuable target samples than existing approaches requiring particular architectures or computation of the distances. Our algorithm, Energy-based Active Domain Adaptation (EADA), queries groups of targe data that incorporate both domain characteristic and instance uncertainty into every selection round. Meanwhile, by aligning the free energy of target data compact around the source domain via a regularization term, domain gap can be implicitly diminished. Through extensive experiments, we show that EADA surpasses state-of-the-art methods on well-known challenging benchmarks with substantial improvements, making it a useful option in the open world. Code is available at //github.com/BIT-DA/EADA.
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.
Object detection typically assumes that training and test data are drawn from an identical distribution, which, however, does not always hold in practice. Such a distribution mismatch will lead to a significant performance drop. In this work, we aim to improve the cross-domain robustness of object detection. We tackle the domain shift on two levels: 1) the image-level shift, such as image style, illumination, etc, and 2) the instance-level shift, such as object appearance, size, etc. We build our approach based on the recent state-of-the-art Faster R-CNN model, and design two domain adaptation components, on image level and instance level, to reduce the domain discrepancy. The two domain adaptation components are based on H-divergence theory, and are implemented by learning a domain classifier in adversarial training manner. The domain classifiers on different levels are further reinforced with a consistency regularization to learn a domain-invariant region proposal network (RPN) in the Faster R-CNN model. We evaluate our newly proposed approach using multiple datasets including Cityscapes, KITTI, SIM10K, etc. The results demonstrate the effectiveness of our proposed approach for robust object detection in various domain shift scenarios.
小貼士
登錄享
注冊地址： 北京市海淀區羊坊店路18號2幢3層301-191