Learning policies which are robust to changes in the environment are critical for real world deployment of Reinforcement Learning agents. They are also necessary for achieving good generalization across environment shifts. We focus on bisimulation metrics, which provide a powerful means for abstracting task relevant components of the observation and learning a succinct representation space for training the agent using reinforcement learning. In this work, we extend the bisimulation framework to also account for context dependent observation shifts. Specifically, we focus on the simulator based learning setting and use alternate observations to learn a representation space which is invariant to observation shifts using a novel bisimulation based objective. This allows us to deploy the agent to varying observation settings during test time and generalize to unseen scenarios. We further provide novel theoretical bounds for simulator fidelity and performance transfer guarantees for using a learnt policy to unseen shifts. Empirical analysis on the high-dimensional image based control domains demonstrates the efficacy of our method.
相關內容
Multi-agent systems require effective coordination between groups and individuals to achieve common goals. However, current multi-agent reinforcement learning (MARL) methods primarily focus on improving individual policies and do not adequately address group-level policies, which leads to weak cooperation. To address this issue, we propose a novel Consensus-oriented Strategy (CoS) that emphasizes group and individual policies simultaneously. Specifically, CoS comprises two main components: (a) the vector quantized group consensus module, which extracts discrete latent embeddings that represent the stable and discriminative group consensus, and (b) the group consensus-oriented strategy, which integrates the group policy using a hypernet and the individual policies using the group consensus, thereby promoting coordination at both the group and individual levels. Through empirical experiments on cooperative navigation tasks with both discrete and continuous spaces, as well as Google research football, we demonstrate that CoS outperforms state-of-the-art MARL algorithms and achieves better collaboration, thus providing a promising solution for achieving effective coordination in multi-agent systems.
We consider estimation and inference with data collected from episodic reinforcement learning (RL) algorithms; i.e. adaptive experimentation algorithms that at each period (aka episode) interact multiple times in a sequential manner with a single treated unit. Our goal is to be able to evaluate counterfactual adaptive policies after data collection and to estimate structural parameters such as dynamic treatment effects, which can be used for credit assignment (e.g. what was the effect of the first period action on the final outcome). Such parameters of interest can be framed as solutions to moment equations, but not minimizers of a population loss function, leading to $Z$-estimation approaches in the case of static data. However, such estimators fail to be asymptotically normal in the case of adaptive data collection. We propose a re-weighted $Z$-estimation approach with carefully designed adaptive weights to stabilize the episode-varying estimation variance, which results from the nonstationary policy that typical episodic RL algorithms invoke. We identify proper weighting schemes to restore the consistency and asymptotic normality of the re-weighted Z-estimators for target parameters, which allows for hypothesis testing and constructing uniform confidence regions for target parameters of interest. Primary applications include dynamic treatment effect estimation and dynamic off-policy evaluation.
Connected automated driving has the potential to significantly improve urban traffic efficiency, e.g., by alleviating issues due to occlusion. Cooperative behavior planning can be employed to jointly optimize the motion of multiple vehicles. Most existing approaches to automatic intersection management, however, only consider fully automated traffic. In practice, mixed traffic, i.e., the simultaneous road usage by automated and human-driven vehicles, will be prevalent. The present work proposes to leverage reinforcement learning and a graph-based scene representation for cooperative multi-agent planning. We build upon our previous works that showed the applicability of such machine learning methods to fully automated traffic. The scene representation is extended for mixed traffic and considers uncertainty in the human drivers' intentions. In the simulation-based evaluation, we model measurement uncertainties through noise processes that are tuned using real-world data. The paper evaluates the proposed method against an enhanced first in - first out scheme, our baseline for mixed traffic management. With increasing share of automated vehicles, the learned planner significantly increases the vehicle throughput and reduces the delay due to interaction. Non-automated vehicles benefit virtually alike.
In this work we propose RELDEC, a novel approach for sequential decoding of moderate length low-density parity-check (LDPC) codes. The main idea behind RELDEC is that an optimized decoding policy is subsequently obtained via reinforcement learning based on a Markov decision process (MDP). In contrast to our previous work, where an agent learns to schedule only a single check node (CN) within a group (cluster) of CNs per iteration, in this work we train the agent to schedule all CNs in a cluster, and all clusters in every iteration. That is, in each learning step of RELDEC an agent learns to schedule CN clusters sequentially depending on a reward associated with the outcome of scheduling a particular cluster. We also modify the state space representation of the MDP, enabling RELDEC to be suitable for larger block length LDPC codes than those studied in our previous work. Furthermore, to address decoding under varying channel conditions, we propose agile meta-RELDEC (AM-RELDEC) that employs meta-reinforcement learning. The proposed RELDEC scheme significantly outperforms standard flooding and random sequential decoding for a variety of LDPC codes, including codes designed for 5G new radio.
Object pushing presents a key non-prehensile manipulation problem that is illustrative of more complex robotic manipulation tasks. While deep reinforcement learning (RL) methods have demonstrated impressive learning capabilities using visual input, a lack of tactile sensing limits their capability for fine and reliable control during manipulation. Here we propose a deep RL approach to object pushing using tactile sensing without visual input, namely tactile pushing. We present a goal-conditioned formulation that allows both model-free and model-based RL to obtain accurate policies for pushing an object to a goal. To achieve real-world performance, we adopt a sim-to-real approach. Our results demonstrate that it is possible to train on a single object and a limited sample of goals to produce precise and reliable policies that can generalize to a variety of unseen objects and pushing scenarios without domain randomization. We experiment with the trained agents in harsh pushing conditions, and show that with significantly more training samples, a model-free policy can outperform a model-based planner, generating shorter and more reliable pushing trajectories despite large disturbances. The simplicity of our training environment and effective real-world performance highlights the value of rich tactile information for fine manipulation. Code and videos are available at //sites.google.com/view/tactile-rl-pushing/.
While Reinforcement Learning (RL) achieves tremendous success in sequential decision-making problems of many domains, it still faces key challenges of data inefficiency and the lack of interpretability. Interestingly, many researchers have leveraged insights from the causality literature recently, bringing forth flourishing works to unify the merits of causality and address well the challenges from RL. As such, it is of great necessity and significance to collate these Causal Reinforcement Learning (CRL) works, offer a review of CRL methods, and investigate the potential functionality from causality toward RL. In particular, we divide existing CRL approaches into two categories according to whether their causality-based information is given in advance or not. We further analyze each category in terms of the formalization of different models, ranging from the Markov Decision Process (MDP), Partially Observed Markov Decision Process (POMDP), Multi-Arm Bandits (MAB), and Dynamic Treatment Regime (DTR). Moreover, we summarize the evaluation matrices and open sources while we discuss emerging applications, along with promising prospects for the future development of CRL.
The transformer architecture and variants presented remarkable success across many machine learning tasks in recent years. This success is intrinsically related to the capability of handling long sequences and the presence of context-dependent weights from the attention mechanism. We argue that these capabilities suit the central role of a Meta-Reinforcement Learning algorithm. Indeed, a meta-RL agent needs to infer the task from a sequence of trajectories. Furthermore, it requires a fast adaptation strategy to adapt its policy for a new task -- which can be achieved using the self-attention mechanism. In this work, we present TrMRL (Transformers for Meta-Reinforcement Learning), a meta-RL agent that mimics the memory reinstatement mechanism using the transformer architecture. It associates the recent past of working memories to build an episodic memory recursively through the transformer layers. We show that the self-attention computes a consensus representation that minimizes the Bayes Risk at each layer and provides meaningful features to compute the best actions. We conducted experiments in high-dimensional continuous control environments for locomotion and dexterous manipulation. Results show that TrMRL presents comparable or superior asymptotic performance, sample efficiency, and out-of-distribution generalization compared to the baselines in these environments.
The dominating NLP paradigm of training a strong neural predictor to perform one task on a specific dataset has led to state-of-the-art performance in a variety of applications (eg. sentiment classification, span-prediction based question answering or machine translation). However, it builds upon the assumption that the data distribution is stationary, ie. that the data is sampled from a fixed distribution both at training and test time. This way of training is inconsistent with how we as humans are able to learn from and operate within a constantly changing stream of information. Moreover, it is ill-adapted to real-world use cases where the data distribution is expected to shift over the course of a model's lifetime. The first goal of this thesis is to characterize the different forms this shift can take in the context of natural language processing, and propose benchmarks and evaluation metrics to measure its effect on current deep learning architectures. We then proceed to take steps to mitigate the effect of distributional shift on NLP models. To this end, we develop methods based on parametric reformulations of the distributionally robust optimization framework. Empirically, we demonstrate that these approaches yield more robust models as demonstrated on a selection of realistic problems. In the third and final part of this thesis, we explore ways of efficiently adapting existing models to new domains or tasks. Our contribution to this topic takes inspiration from information geometry to derive a new gradient update rule which alleviate catastrophic forgetting issues during adaptation.
Recent advances in representation learning have demonstrated an ability to represent information from different modalities such as video, text, and audio in a single high-level embedding vector. In this work we present a self-supervised learning framework that is able to learn a representation that captures finer levels of granularity across different modalities such as concepts or events represented by visual objects or spoken words. Our framework relies on a discretized embedding space created via vector quantization that is shared across different modalities. Beyond the shared embedding space, we propose a Cross-Modal Code Matching objective that forces the representations from different views (modalities) to have a similar distribution over the discrete embedding space such that cross-modal objects/actions localization can be performed without direct supervision. In our experiments we show that the proposed discretized multi-modal fine-grained representation (e.g., pixel/word/frame) can complement high-level summary representations (e.g., video/sentence/waveform) for improved performance on cross-modal retrieval tasks. We also observe that the discretized representation uses individual clusters to represent the same semantic concept across modalities.
Reinforcement learning (RL) is a popular paradigm for addressing sequential decision tasks in which the agent has only limited environmental feedback. Despite many advances over the past three decades, learning in many domains still requires a large amount of interaction with the environment, which can be prohibitively expensive in realistic scenarios. To address this problem, transfer learning has been applied to reinforcement learning such that experience gained in one task can be leveraged when starting to learn the next, harder task. More recently, several lines of research have explored how tasks, or data samples themselves, can be sequenced into a curriculum for the purpose of learning a problem that may otherwise be too difficult to learn from scratch. In this article, we present a framework for curriculum learning (CL) in reinforcement learning, and use it to survey and classify existing CL methods in terms of their assumptions, capabilities, and goals. Finally, we use our framework to find open problems and suggest directions for future RL curriculum learning research.
注冊地址： 北京市海淀區羊坊店路18號2幢3層301-191