Recently, model-based reinforcement learning algorithms have demonstrated remarkable efficacy in visual input environments. These approaches begin by constructing a parameterized simulation world model of the real environment through self-supervised learning. By leveraging the imagination of the world model, the agent's policy is enhanced without the constraints of sampling from the real environment. The performance of these algorithms heavily relies on the sequence modeling and generation capabilities of the world model. However, constructing a perfectly accurate model of a complex unknown environment is nearly impossible. Discrepancies between the model and reality may cause the agent to pursue virtual goals, resulting in subpar performance in the real environment. Introducing random noise into model-based reinforcement learning has been proven beneficial. In this work, we introduce Stochastic Transformer-based wORld Model (STORM), an efficient world model architecture that combines the strong sequence modeling and generation capabilities of Transformers with the stochastic nature of variational autoencoders. STORM achieves a mean human performance of $126.7\%$ on the Atari $100$k benchmark, setting a new record among state-of-the-art methods that do not employ lookahead search techniques. Moreover, training an agent with $1.85$ hours of real-time interaction experience on a single NVIDIA GeForce RTX 3090 graphics card requires only $4.3$ hours, showcasing improved efficiency compared to previous methodologies.
Motivated by the challenge of achieving rapid learning in physical environments, this paper presents the development and training of a robotic system designed to navigate and solve a labyrinth game using model-based reinforcement learning techniques. The method involves extracting low-dimensional observations from camera images, along with a cropped and rectified image patch centered on the current position within the labyrinth, providing valuable information about the labyrinth layout. The learning of a control policy is performed purely on the physical system using model-based reinforcement learning, where the progress along the labyrinth's path serves as a reward signal. Additionally, we exploit the system's inherent symmetries to augment the training data. Consequently, our approach learns to successfully solve a popular real-world labyrinth game in record time, with only 5 hours of real-world training data.
We propose a Gaussian manifold variational auto-encoder (GM-VAE) whose latent space consists of a set of Gaussian distributions. It is known that the set of the univariate Gaussian distributions with the Fisher information metric form a hyperbolic space, which we call a Gaussian manifold. To learn the VAE endowed with the Gaussian manifolds, we propose a pseudo-Gaussian manifold normal distribution based on the Kullback-Leibler divergence, a local approximation of the squared Fisher-Rao distance, to define a density over the latent space. In experiments, we demonstrate the efficacy of GM-VAE on two different tasks: density estimation of image datasets and environment modeling in model-based reinforcement learning. GM-VAE outperforms the other variants of hyperbolic- and Euclidean-VAEs on density estimation tasks and shows competitive performance in model-based reinforcement learning. We observe that our model provides strong numerical stability, addressing a common limitation reported in previous hyperbolic-VAEs.
In the backdrop of an increasingly pressing need for effective urban and highway transportation systems, this work explores the synergy between model-based and learning-based strategies to enhance traffic flow management by use of an innovative approach to the problem of ramp metering control that embeds Reinforcement Learning (RL) techniques within the Model Predictive Control (MPC) framework. The control problem is formulated as an RL task by crafting a suitable stage cost function that is representative of the traffic conditions, variability in the control action, and violations of the constraint on the maximum number of vehicles in queue. An MPC-based RL approach, which leverages the MPC optimal problem as a function approximation for the RL algorithm, is proposed to learn to efficiently control an on-ramp and satisfy its constraints despite uncertainties in the system model and variable demands. Simulations are performed on a benchmark small-scale highway network to compare the proposed methodology against other state-of-the-art control approaches. Results show that, starting from an MPC controller that has an imprecise model and is poorly tuned, the proposed methodology is able to effectively learn to improve the control policy such that congestion in the network is reduced and constraints are satisfied, yielding an improved performance that is superior to the other controllers.
We present a novel framework for generating adversarial benchmarks to evaluate the robustness of image classification models. Our framework allows users to customize the types of distortions to be optimally applied to images, which helps address the specific distortions relevant to their deployment. The benchmark can generate datasets at various distortion levels to assess the robustness of different image classifiers. Our results show that the adversarial samples generated by our framework with any of the image classification models, like ResNet-50, Inception-V3, and VGG-16, are effective and transferable to other models causing them to fail. These failures happen even when these models are adversarially retrained using state-of-the-art techniques, demonstrating the generalizability of our adversarial samples. We achieve competitive performance in terms of net $L_2$ distortion compared to state-of-the-art benchmark techniques on CIFAR-10 and ImageNet; however, we demonstrate our framework achieves such results with simple distortions like Gaussian noise without introducing unnatural artifacts or color bleeds. This is made possible by a model-based reinforcement learning (RL) agent and a technique that reduces a deep tree search of the image for model sensitivity to perturbations, to a one-level analysis and action. The flexibility of choosing distortions and setting classification probability thresholds for multiple classes makes our framework suitable for algorithmic audits.
Model-based Reinforcement Learning (MBRL) aims to make agents more sample-efficient, adaptive, and explainable by learning an explicit model of the environment. While the capabilities of MBRL agents have significantly improved in recent years, how to best learn the model is still an unresolved question. The majority of MBRL algorithms aim at training the model to make accurate predictions about the environment and subsequently using the model to determine the most rewarding actions. However, recent research has shown that model predictive accuracy is often not correlated with action quality, tracing the root cause to the objective mismatch between accurate dynamics model learning and policy optimization of rewards. A number of interrelated solution categories to the objective mismatch problem have emerged as MBRL continues to mature as a research area. In this work, we provide an in-depth survey of these solution categories and propose a taxonomy to foster future research.
While language models (LMs) have shown potential across a range of decision-making tasks, their reliance on simple acting processes limits their broad deployment as autonomous agents. In this paper, we introduce Language Agent Tree Search (LATS) -- the first general framework that synergizes the capabilities of LMs in reasoning, acting, and planning. By leveraging the in-context learning ability of LMs, we integrate Monte Carlo Tree Search into LATS to enable LMs as agents, along with LM-powered value functions and self-reflections for proficient exploration and enhanced decision-making. A key feature of our approach is the incorporation of an environment for external feedback, which offers a more deliberate and adaptive problem-solving mechanism that surpasses the constraints of existing techniques. Our experimental evaluation across diverse domains, including programming, interactive question-answering (QA), web navigation, and math, validates the effectiveness and generality of LATS in decision-making while maintaining competitive or improved reasoning performance. Notably, LATS achieves state-of-the-art pass@1 accuracy (92.7%) for programming on HumanEval with GPT-4 and demonstrates gradient-free performance (average score of 75.9) comparable to gradient-based fine-tuning for web navigation on WebShop with GPT-3.5. Code can be found at https://github.com/lapisrocks/LanguageAgentTreeSearch
Model-based reinforcement learning seeks to simultaneously learn the dynamics of an unknown stochastic environment and synthesise an optimal policy for acting in it. Ensuring the safety and robustness of sequential decisions made through a policy in such an environment is a key challenge for policies intended for safety-critical scenarios. In this work, we investigate two complementary problems: first, computing reach-avoid probabilities for iterative predictions made with dynamical models, with dynamics described by Bayesian neural network (BNN); second, synthesising control policies that are optimal with respect to a given reach-avoid specification (reaching a "target" state, while avoiding a set of "unsafe" states) and a learned BNN model. Our solution leverages interval propagation and backward recursion techniques to compute lower bounds for the probability that a policy's sequence of actions leads to satisfying the reach-avoid specification. Such computed lower bounds provide safety certification for the given policy and BNN model. We then introduce control synthesis algorithms to derive policies maximizing said lower bounds on the safety probability. We demonstrate the effectiveness of our method on a series of control benchmarks characterized by learned BNN dynamics models. On our most challenging benchmark, compared to purely data-driven policies the optimal synthesis algorithm is able to provide more than a four-fold increase in the number of certifiable states and more than a three-fold increase in the average guaranteed reach-avoid probability.
Model-based reinforcement learning (MBRL) holds the promise of sample-efficient learning by utilizing a world model, which models how the environment works and typically encompasses components for two tasks: observation modeling and reward modeling. In this paper, through a dedicated empirical investigation, we gain a deeper understanding of the role each task plays in world models and uncover the overlooked potential of sample-efficient MBRL by mitigating the domination of either observation or reward modeling. Our key insight is that while prevalent approaches of explicit MBRL attempt to restore abundant details of the environment via observation models, it is difficult due to the environment's complexity and limited model capacity. On the other hand, reward models, while dominating implicit MBRL and adept at learning compact task-centric dynamics, are inadequate for sample-efficient learning without richer learning signals. Motivated by these insights and discoveries, we propose a simple yet effective approach, HarmonyDream, which automatically adjusts loss coefficients to maintain task harmonization, i.e. a dynamic equilibrium between the two tasks in world model learning. Our experiments show that the base MBRL method equipped with HarmonyDream gains 10%-69% absolute performance boosts on visual robotic tasks and sets a new state-of-the-art result on the Atari 100K benchmark. Code is available at https://github.com/thuml/HarmonyDream.
Inspired by human conscious planning, we propose Skipper, a model-based reinforcement learning framework utilizing spatio-temporal abstractions to generalize better in novel situations. It automatically decomposes the given task into smaller, more manageable subtasks, and thus enables sparse decision-making and focused computation on the relevant parts of the environment. The decomposition relies on the extraction of an abstracted proxy problem represented as a directed graph, in which vertices and edges are learned end-to-end from hindsight. Our theoretical analyses provide performance guarantees under appropriate assumptions and establish where our approach is expected to be helpful. Generalization-focused experiments validate Skipper's significant advantage in zero-shot generalization, compared to some existing state-of-the-art hierarchical planning methods.
This paper addresses the prediction stability, prediction accuracy and control capability of the current probabilistic model-based reinforcement learning (MBRL) built on neural networks. A novel approach dropout-based probabilistic ensembles with trajectory sampling (DPETS) is proposed where the system uncertainty is stably predicted by combining the Monte-Carlo dropout and trajectory sampling in one framework. Its loss function is designed to correct the fitting error of neural networks for more accurate prediction of probabilistic models. The state propagation in its policy is extended to filter the aleatoric uncertainty for superior control capability. Evaluated by several Mujoco benchmark control tasks under additional disturbances and one practical robot arm manipulation task, DPETS outperforms related MBRL approaches in both average return and convergence velocity while achieving superior performance than well-known model-free baselines with significant sample efficiency. The open source code of DPETS is available at https://github.com/mrjun123/DPETS.
Due to the proliferation of malware, defenders are increasingly turning to automation and machine learning as part of the malware detection tool-chain. However, machine learning models are susceptible to adversarial attacks, requiring the testing of model and product robustness. Meanwhile, attackers also seek to automate malware generation and evasion of antivirus systems, and defenders try to gain insight into their methods. This work proposes a new algorithm that combines Malware Evasion and Model Extraction (MEME) attacks. MEME uses model-based reinforcement learning to adversarially modify Windows executable binary samples while simultaneously training a surrogate model with a high agreement with the target model to evade. To evaluate this method, we compare it with two state-of-the-art attacks in adversarial malware creation, using three well-known published models and one antivirus product as targets. Results show that MEME outperforms the state-of-the-art methods in terms of evasion capabilities in almost all cases, producing evasive malware with an evasion rate in the range of 32-73%. It also produces surrogate models with a prediction label agreement with the respective target models between 97-99%. The surrogate could be used to fine-tune and improve the evasion rate in the future.
Quantifying uncertainty about a policy's long-term performance is important to solve sequential decision-making tasks. We study the problem from a model-based Bayesian reinforcement learning perspective, where the goal is to learn the posterior distribution over value functions induced by parameter (epistemic) uncertainty of the Markov decision process. Previous work restricts the analysis to a few moments of the distribution over values or imposes a particular distribution shape, e.g., Gaussians. Inspired by distributional reinforcement learning, we introduce a Bellman operator whose fixed-point is the value distribution function. Based on our theory, we propose Epistemic Quantile-Regression (EQR), a model-based algorithm that learns a value distribution function. We combine EQR with soft actor-critic (SAC) for policy optimization with an arbitrary differentiable objective function of the learned value distribution. Evaluation across several continuous-control tasks shows performance benefits with respect to both model-based and model-free algorithms. The code is available at https://github.com/boschresearch/dist-mbrl.
Recent successes in image generation, model-based reinforcement learning, and text-to-image generation have demonstrated the empirical advantages of discrete latent representations, although the reasons behind their benefits remain unclear. We explore the relationship between discrete latent spaces and disentangled representations by replacing the standard Gaussian variational autoencoder (VAE) with a tailored categorical variational autoencoder. We show that the underlying grid structure of categorical distributions mitigates the problem of rotational invariance associated with multivariate Gaussian distributions, acting as an efficient inductive prior for disentangled representations. We provide both analytical and empirical findings that demonstrate the advantages of discrete VAEs for learning disentangled representations. Furthermore, we introduce the first unsupervised model selection strategy that favors disentangled representations.
Agents must be able to adapt quickly as an environment changes. We find that existing model-based reinforcement learning agents are unable to do this well, in part because of how they use past experiences to train their world model. Here, we present Curious Replay -- a form of prioritized experience replay tailored to model-based agents through use of a curiosity-based priority signal. Agents using Curious Replay exhibit improved performance in an exploration paradigm inspired by animal behavior and on the Crafter benchmark. DreamerV3 with Curious Replay surpasses state-of-the-art performance on Crafter, achieving a mean score of 19.4 that substantially improves on the previous high score of 14.5 by DreamerV3 with uniform replay, while also maintaining similar performance on the Deepmind Control Suite. Code for Curious Replay is available at https://github.com/AutonomousAgentsLab/curiousreplay
Pretraining reinforcement learning (RL) models on offline datasets is a promising way to improve their training efficiency in online tasks, but challenging due to the inherent mismatch in dynamics and behaviors across various tasks. We present a model-based RL method that learns to transfer potentially useful dynamics and action demonstrations from offline data to a novel task. The main idea is to use the world models not only as simulators for behavior learning but also as tools to measure the task relevance for both dynamics representation transfer and policy transfer. We build a time-varying, domain-selective distillation loss to generate a set of offline-to-online similarity weights. These weights serve two purposes: (i) adaptively transferring the task-agnostic knowledge of physical dynamics to facilitate world model training, and (ii) learning to replay relevant source actions to guide the target policy. We demonstrate the advantages of our approach compared with the state-of-the-art methods in Meta-World and DeepMind Control Suite.
Local-remote systems allow robots to execute complex tasks in hazardous environments such as space and nuclear power stations. However, establishing accurate positional mapping between local and remote devices can be difficult due to time delays that can compromise system performance and stability. Enhancing the synchronicity and stability of local-remote systems is vital for enabling robots to interact with environments at greater distances and under highly challenging network conditions, including time delays. We introduce an adaptive control method employing reinforcement learning to tackle the time-delayed control problem. By adjusting controller parameters in real-time, this adaptive controller compensates for stochastic delays and improves synchronicity between local and remote robotic manipulators. To improve the adaptive PD controller's performance, we devise a model-based reinforcement learning approach that effectively incorporates multi-step delays into the learning framework. Utilizing this proposed technique, the local-remote system's performance is stabilized for stochastic communication time-delays of up to 290ms. Our results demonstrate that the suggested model-based reinforcement learning method surpasses the Soft-Actor Critic and augmented state Soft-Actor Critic techniques. Access the code at: https://github.com/CAV-Research-Lab/Predictive-Model-Delay-Correction
We propose a novel generalization of constrained Markov decision processes (CMDPs) that we call the \emph{semi-infinitely constrained Markov decision process} (SICMDP). Particularly, we consider a continuum of constraints instead of a finite number of constraints as in the case of ordinary CMDPs. We also devise two reinforcement learning algorithms for SICMDPs that we call SI-CRL and SI-CPO. SI-CRL is a model-based reinforcement learning algorithm. Given an estimate of the transition model, we first transform the reinforcement learning problem into a linear semi-infinitely programming (LSIP) problem and then use the dual exchange method in the LSIP literature to solve it. SI-CPO is a policy optimization algorithm. Borrowing the ideas from the cooperative stochastic approximation approach, we make alternative updates to the policy parameters to maximize the reward or minimize the cost. To the best of our knowledge, we are the first to apply tools from semi-infinitely programming (SIP) to solve constrained reinforcement learning problems. We present theoretical analysis for SI-CRL and SI-CPO, identifying their iteration complexity and sample complexity. We also conduct extensive numerical examples to illustrate the SICMDP model and demonstrate that our proposed algorithms are able to solve complex sequential decision-making tasks leveraging modern deep reinforcement learning techniques.
Despite remarkable successes, deep reinforcement learning algorithms remain sample inefficient: they require an enormous amount of trial and error to find good policies. Model-based algorithms promise sample efficiency by building an environment model that can be used for planning. Posterior Sampling for Reinforcement Learning is such a model-based algorithm that has attracted significant interest due to its performance in the tabular setting. This paper introduces Posterior Sampling for Deep Reinforcement Learning (PSDRL), the first truly scalable approximation of Posterior Sampling for Reinforcement Learning that retains its model-based essence. PSDRL combines efficient uncertainty quantification over latent state space models with a specially tailored continual planning algorithm based on value-function approximation. Extensive experiments on the Atari benchmark show that PSDRL significantly outperforms previous state-of-the-art attempts at scaling up posterior sampling while being competitive with a state-of-the-art (model-based) reinforcement learning method, both in sample efficiency and computational efficiency.
Model-based reinforcement learning is a powerful tool, but collecting data to fit an accurate model of the system can be costly. Exploring an unknown environment in a sample-efficient manner is hence of great importance. However, the complexity of dynamics and the computational limitations of real systems make this task challenging. In this work, we introduce FLEX, an exploration algorithm for nonlinear dynamics based on optimal experimental design. Our policy maximizes the information of the next step and results in an adaptive exploration algorithm, compatible with generic parametric learning models and requiring minimal resources. We test our method on a number of nonlinear environments covering different settings, including time-varying dynamics. Keeping in mind that exploration is intended to serve an exploitation objective, we also test our algorithm on downstream model-based classical control tasks and compare it to other state-of-the-art model-based and model-free approaches. The performance achieved by FLEX is competitive and its computational cost is low.