Emergence and causality are two fundamental concepts for understanding complex systems. They are interconnected. On one hand, emergence refers to the phenomenon where macroscopic properties cannot be solely attributed to the cause of individual properties. On the other hand, causality can exhibit emergence, meaning that new causal laws may arise as we increase the level of abstraction. Causal emergence theory aims to bridge these two concepts and even employs measures of causality to quantify emergence. This paper provides a comprehensive review of recent advancements in quantitative theories and applications of causal emergence. Two key problems are addressed: quantifying causal emergence and identifying it in data. Addressing the latter requires the use of machine learning techniques, thus establishing a connection between causal emergence and artificial intelligence. We highlighted that the architectures used for identifying causal emergence are shared by causal representation learning, causal model abstraction, and world model-based reinforcement learning. Consequently, progress in any of these areas can benefit the others. Potential applications and future perspectives are also discussed in the final section of the review.
This article proposes an improved trajectory optimization approach for stochastic optimal control of dynamical systems affected by measurement noise by combining optimal control with maximum likelihood techniques to improve the reduction of the cumulative cost-to-go. A modified optimization objective function that incorporates dynamic programming-based controller design is presented to handle the noise in the system and sensors. Empirical results demonstrate the effectiveness of the approach in reducing stochasticity and allowing for an intermediate step to switch optimization that can allow an efficient balance of exploration and exploitation mechanism for complex tasks by constraining policy parameters to parameters obtained as a result of this improved optimization. This research study also includes theoretical work on the uniqueness of control parameter estimates and also leverages a structure of the likelihood function which has an established theoretical guarantees. Furthermore, a theoretical result is also explored that bridge the gap between the proposed optimization objective function and existing information theory (relative entropy) and optimal control dualities.
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.
Autonomous shipping has recently gained much interest in the research community. However, little research focuses on inland - and port navigation, even though this is identified by countries such as Belgium and the Netherlands as an essential step towards a sustainable future. These environments pose unique challenges, since they can contain dynamic obstacles that do not broadcast their location, such as small vessels, kayaks or buoys. Therefore, this research proposes a navigational algorithm which can navigate an inland vessel in a wide variety of complex port scenarios using ranging sensors to observe the environment. The proposed methodology is based on a machine learning approach that has recently set benchmark results in various domains: model-based reinforcement learning. By randomizing the port environments during training, the trained model can navigate in scenarios that it never encountered during training. Furthermore, results show that our approach outperforms the commonly used dynamic window approach and a benchmark model-free reinforcement learning algorithm. This work is therefore a significant step towards vessels that can navigate autonomously in complex port scenarios.
We consider the problem of quantifying uncertainty over expected cumulative rewards in model-based reinforcement learning. In particular, we focus on characterizing the variance over values induced by a distribution over Markov decision processes (MDPs). Previous work upper bounds the posterior variance over values by solving a so-called uncertainty Bellman equation (UBE), but the over-approximation may result in inefficient exploration. We propose a new UBE whose solution converges to the true posterior variance over values and leads to lower regret in tabular exploration problems. We identify challenges to apply the UBE theory beyond tabular problems and propose a suitable approximation. Based on this approximation, we introduce a general-purpose policy optimization algorithm, Q-Uncertainty Soft Actor-Critic (QU-SAC), that can be applied for either risk-seeking or risk-averse policy optimization with minimal changes. Experiments in both online and offline RL demonstrate improved performance compared to other uncertainty estimation methods.
In this paper, we introduce a policy-gradient method for model-based reinforcement learning (RL) that exploits a type of stationary distributions commonly obtained from Markov decision processes (MDPs) in stochastic networks, queueing systems, and statistical mechanics. Specifically, when the stationary distribution of the MDP belongs to an exponential family that is parametrized by policy parameters, we can improve existing policy gradient methods for average-reward RL. Our key identification is a family of gradient estimators, called score-aware gradient estimators (SAGEs), that enable policy gradient estimation without relying on value-function approximation in the aforementioned setting. This contrasts with other common policy-gradient algorithms such as actor-critic methods. We first show that policy-gradient with SAGE locally converges, including in cases when the objective function is nonconvex, presents multiple maximizers, and the state space of the MDP is not finite. Under appropriate assumptions such as starting sufficiently close to a maximizer, the policy under stochastic gradient ascent with SAGE has an overwhelming probability of converging to the associated optimal policy. Other key assumptions are that a local Lyapunov function exists, and a nondegeneracy property of the Hessian of the objective function holds locally around a maximizer. Furthermore, we conduct a numerical comparison between a SAGE-based policy-gradient method and an actor-critic method. We specifically focus on several examples inspired from stochastic networks, queueing systems, and models derived from statistical physics, where parametrizable exponential families are commonplace. Our results demonstrate that a SAGE-based method finds close-to-optimal policies faster than an actor-critic method.
We propose regularity as a novel reward signal for intrinsically-motivated reinforcement learning. Taking inspiration from child development, we postulate that striving for structure and order helps guide exploration towards a subspace of tasks that are not favored by naive uncertainty-based intrinsic rewards. Our generalized formulation of Regularity as Intrinsic Reward (RaIR) allows us to operationalize it within model-based reinforcement learning. In a synthetic environment, we showcase the plethora of structured patterns that can emerge from pursuing this regularity objective. We also demonstrate the strength of our method in a multi-object robotic manipulation environment. We incorporate RaIR into free play and use it to complement the model's epistemic uncertainty as an intrinsic reward. Doing so, we witness the autonomous construction of towers and other regular structures during free play, which leads to a substantial improvement in zero-shot downstream task performance on assembly tasks.
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.
Recent advances in deep reinforcement learning have showcased its potential in tackling complex tasks. However, experiments on visual control tasks have revealed that state-of-the-art reinforcement learning models struggle with out-of-distribution generalization. Conversely, expressing higher-level concepts and global contexts is relatively easy using language. Building upon recent success of the large language models, our main objective is to improve the state abstraction technique in reinforcement learning by leveraging language for robust action selection. Specifically, we focus on learning language-grounded visual features to enhance the world model learning, a model-based reinforcement learning technique. To enforce our hypothesis explicitly, we mask out the bounding boxes of a few objects in the image observation and provide the text prompt as descriptions for these masked objects. Subsequently, we predict the masked objects along with the surrounding regions as pixel reconstruction, similar to the transformer-based masked autoencoder approach. Our proposed LanGWM: Language Grounded World Model achieves state-of-the-art performance in out-of-distribution test at the 100K interaction steps benchmarks of iGibson point navigation tasks. Furthermore, our proposed technique of explicit language-grounded visual representation learning has the potential to improve models for human-robot interaction because our extracted visual features are language grounded.
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.
Reinforcement learning (RL) has a rich history in neuroscience, from early work on dopamine as a reward prediction error signal (Schultz et al., 1997) to recent work proposing that the brain could implement a form of 'distributional reinforcement learning' popularized in machine learning (Dabney et al., 2020). There has been a close link between theoretical advances in reinforcement learning and neuroscience experiments throughout this literature, and the theories describing the experimental data have therefore become increasingly complex. Here, we provide an introduction and mathematical background to many of the methods that have been used in systems neroscience. We start with an overview of the RL problem and classical temporal difference algorithms, followed by a discussion of 'model-free', 'model-based', and intermediate RL algorithms. We then introduce deep reinforcement learning and discuss how this framework has led to new insights in neuroscience. This includes a particular focus on meta-reinforcement learning (Wang et al., 2018) and distributional RL (Dabney et al., 2020). Finally, we discuss potential shortcomings of the RL formalism for neuroscience and highlight open questions in the field. Code that implements the methods discussed and generates the figures is also provided.
Digital twins promise to revolutionize engineering by offering new avenues for optimization, control, and predictive maintenance. We propose a novel framework for simultaneously training the digital twin of an engineering system and an associated control agent. The twin's training combines adjoint-based data assimilation and system identification methods, while the control agent's training merges model-based optimal control with model-free reinforcement learning. The control agent evolves along two independent paths: one driven by model-based optimal control and the other by reinforcement learning. The digital twin serves as a virtual environment for confrontation and indirect interaction, functioning as an "expert demonstrator." The best policy is selected for real-world interaction and cloned to the other path if training stagnates. We call this framework Reinforcement Twinning (RT). The framework is tested on three diverse engineering systems and control tasks: (1) controlling a wind turbine under varying wind speeds, (2) trajectory control of flapping-wing micro air vehicles (FWMAVs) facing wind gusts, and (3) mitigating thermal loads in managing cryogenic storage tanks. These test cases use simplified models with known ground truth closure laws. Results show that the adjoint-based digital twin training is highly sample-efficient, completing within a few iterations. For the control agent training, both model-based and model-free approaches benefit from their complementary learning experiences. The promising results pave the way for implementing the RT framework on real systems.
Model-based reinforcement learning (MBRL) has gained much attention for its ability to learn complex behaviors in a sample-efficient way: planning actions by generating imaginary trajectories with predicted rewards. Despite its success, we found that surprisingly, reward prediction is often a bottleneck of MBRL, especially for sparse rewards that are challenging (or even ambiguous) to predict. Motivated by the intuition that humans can learn from rough reward estimates, we propose a simple yet effective reward smoothing approach, DreamSmooth, which learns to predict a temporally-smoothed reward, instead of the exact reward at the given timestep. We empirically show that DreamSmooth achieves state-of-the-art performance on long-horizon sparse-reward tasks both in sample efficiency and final performance without losing performance on common benchmarks, such as Deepmind Control Suite and Atari benchmarks.
Reinforcement learning from human feedback (RLHF) has emerged as a powerful technique to make large language models (LLMs) more capable in complex settings. RLHF proceeds as collecting human preference data, training a reward model on said data, and optimizing a base ML model with respect to said reward for extrinsic evaluation metrics (e.g. MMLU, GSM8k). RLHF relies on many assumptions about how the various pieces fit together, such as a reward model capturing human preferences and an RL optimizer extracting the right signal from a reward model. As the RLHF process involves many distinct design decisions, it is easy to assume that multiple processes are correlated and therefore numerically linked. This apparent correlation is often not true, where reward models are easily overoptimized or RL optimizers can reduce performance on tasks not modeled in the data. Notable manifestations of models trained with imperfect RLHF systems are those that are prone to refusing basic requests for safety reasons or appearing lazy in generations. As chat model evaluation becomes increasingly nuanced, the reliance on a perceived link between reward model training, RL scores, and downstream performance drives these issues, which we describe as an objective mismatch. In this paper, we illustrate the causes of this issue, reviewing relevant literature from model-based reinforcement learning, and argue for solutions. By solving objective mismatch in RLHF, the ML models of the future will be more precisely aligned to user instructions for both safety and helpfulness.
Reinforcement learning algorithms typically consider discrete-time dynamics, even though the underlying systems are often continuous in time. In this paper, we introduce a model-based reinforcement learning algorithm that represents continuous-time dynamics using nonlinear ordinary differential equations (ODEs). We capture epistemic uncertainty using well-calibrated probabilistic models, and use the optimistic principle for exploration. Our regret bounds surface the importance of the measurement selection strategy(MSS), since in continuous time we not only must decide how to explore, but also when to observe the underlying system. Our analysis demonstrates that the regret is sublinear when modeling ODEs with Gaussian Processes (GP) for common choices of MSS, such as equidistant sampling. Additionally, we propose an adaptive, data-dependent, practical MSS that, when combined with GP dynamics, also achieves sublinear regret with significantly fewer samples. We showcase the benefits of continuous-time modeling over its discrete-time counterpart, as well as our proposed adaptive MSS over standard baselines, on several applications.
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.
The advances in unsupervised object-centric representation learning have significantly improved its application to downstream tasks. Recent works highlight that disentangled object representations can aid policy learning in image-based, object-centric reinforcement learning tasks. This paper proposes a novel object-centric reinforcement learning algorithm that integrates actor-critic and model-based approaches by incorporating an object-centric world model within the critic. The world model captures the environment's data-generating process by predicting the next state and reward given the current state-action pair, where actions are interventions in the environment. In model-based reinforcement learning, world model learning can be interpreted as a causal induction problem, where the agent must learn the causal relationships underlying the environment's dynamics. We evaluate our method in a simulated 3D robotic environment and a 2D environment with compositional structure. As baselines, we compare against object-centric, model-free actor-critic algorithms and a state-of-the-art monolithic model-based algorithm. While the baselines show comparable performance in easier tasks, our approach outperforms them in more challenging scenarios with a large number of objects or more complex dynamics.
TD-MPC is a model-based reinforcement learning (RL) algorithm that performs local trajectory optimization in the latent space of a learned implicit (decoder-free) world model. In this work, we present TD-MPC2: a series of improvements upon the TD-MPC algorithm. We demonstrate that TD-MPC2 improves significantly over baselines across 104 online RL tasks spanning 4 diverse task domains, achieving consistently strong results with a single set of hyperparameters. We further show that agent capabilities increase with model and data size, and successfully train a single 317M parameter agent to perform 80 tasks across multiple task domains, embodiments, and action spaces. We conclude with an account of lessons, opportunities, and risks associated with large TD-MPC2 agents. Explore videos, models, data, code, and more at https://tdmpc2.com
The primacy bias in model-free reinforcement learning (MFRL), which refers to the agent's tendency to overfit early data and lose the ability to learn from new data, can significantly decrease the performance of MFRL algorithms. Previous studies have shown that employing simple techniques, such as resetting the agent's parameters, can substantially alleviate the primacy bias in MFRL. However, the primacy bias in model-based reinforcement learning (MBRL) remains unexplored. In this work, we focus on investigating the primacy bias in MBRL. We begin by observing that resetting the agent's parameters harms its performance in the context of MBRL. We further find that the primacy bias in MBRL is more closely related to the primacy bias of the world model instead of the primacy bias of the agent. Based on this finding, we propose \textit{world model resetting}, a simple yet effective technique to alleviate the primacy bias in MBRL. We apply our method to two different MBRL algorithms, MBPO and DreamerV2. We validate the effectiveness of our method on multiple continuous control tasks on MuJoCo and DeepMind Control Suite, as well as discrete control tasks on Atari 100k benchmark. The experimental results show that \textit{world model resetting} can significantly alleviate the primacy bias in the model-based setting and improve the algorithm's performance. We also give a guide on how to perform \textit{world model resetting} effectively.