We propose and theoretically analyze an approach for planning with an approximate model in reinforcement learning that can reduce the adverse impact of model error. If the model is accurate enough, it accelerates the convergence to the true value function too. One of its key components is the MaxEnt Model Correction (MoCo) procedure that corrects the model's next-state distributions based on a Maximum Entropy density estimation formulation. Based on MoCo, we introduce the Model Correcting Value Iteration (MoCoVI) algorithm, and its sampled-based variant MoCoDyna. We show that MoCoVI and MoCoDyna's convergence can be much faster than the conventional model-free algorithms. Unlike traditional model-based algorithms, MoCoVI and MoCoDyna effectively utilize an approximate model and still converge to the correct value function.

Why do Recurrent State Space Models such as PlaNet fail at cloth manipulation tasks? Recent work has attributed this to the blurry prediction of the observation, which makes it difficult to plan directly in the latent space. This paper explores the reasons behind this by applying PlaNet in the pick-and-place fabric-flattening domain. We find that the sharp discontinuity of the transition function on the contour of the fabric makes it difficult to learn an accurate latent dynamic model, causing the MPC planner to produce pick actions slightly outside of the article. By limiting picking space on the cloth mask and training on specially engineered trajectories, our mesh-free PlaNet-ClothPick surpasses visual planning and policy learning methods on principal metrics in simulation, achieving similar performance as state-of-the-art mesh-based planning approaches. Notably, our model exhibits a faster action inference and requires fewer transitional model parameters than the state-of-the-art robotic systems in this domain. Other supplementary materials are available at: https://sites.google.com/view/planet-clothpick.

Diffusion models are the current state of the art for generating photorealistic images. Controlling the sampling process for constrained image generation tasks such as inpainting, however, remains challenging since exact conditioning on such constraints is intractable. While existing methods use various techniques to approximate the constrained posterior, this paper proposes to exploit the ability of Tractable Probabilistic Models (TPMs) to exactly and efficiently compute the constrained posterior, and to leverage this signal to steer the denoising process of diffusion models. Specifically, this paper adopts a class of expressive TPMs termed Probabilistic Circuits (PCs). Building upon prior advances, we further scale up PCs and make them capable of guiding the image generation process of diffusion models. Empirical results suggest that our approach can consistently improve the overall quality and semantic coherence of inpainted images across three natural image datasets (i.e., CelebA-HQ, ImageNet, and LSUN) with only ~10% additional computational overhead brought by the TPM. Further, with the help of an image encoder and decoder, our method can readily accept semantic constraints on specific regions of the image, which opens up the potential for more controlled image generation tasks. In addition to proposing a new framework for constrained image generation, this paper highlights the benefit of more tractable models and motivates the development of expressive TPMs.

Large language models (LLMs) solve problems more accurately and interpretably when instructed to work out the answer step by step using a ``chain-of-thought'' (CoT) prompt. One can also improve LLMs' performance on a specific task by supervised fine-tuning, i.e., by using gradient ascent on some tunable parameters to maximize the average log-likelihood of correct answers from a labeled training set. Naively combining CoT with supervised tuning requires supervision not just of the correct answers, but also of detailed rationales that lead to those answers; these rationales are expensive to produce by hand. Instead, we propose a fine-tuning strategy that tries to maximize the \emph{marginal} log-likelihood of generating a correct answer using CoT prompting, approximately averaging over all possible rationales. The core challenge is sampling from the posterior over rationales conditioned on the correct answer; we address it using a simple Markov-chain Monte Carlo (MCMC) expectation-maximization (EM) algorithm inspired by the self-taught reasoner (STaR), memoized wake-sleep, Markovian score climbing, and persistent contrastive divergence. This algorithm also admits a novel control-variate technique that drives the variance of our gradient estimates to zero as the model improves. Applying our technique to GSM8K and the tasks in BIG-Bench Hard, we find that this MCMC-EM fine-tuning technique typically improves the model's accuracy on held-out examples more than STaR or prompt-tuning with or without CoT.

Optimally scheduling multi-energy flow is an effective method to utilize renewable energy sources (RES) and improve the stability and economy of integrated energy systems (IES). However, the stable demand-supply of IES faces challenges from uncertainties that arise from RES and loads, as well as the increasing impact of cyber-attacks with advanced information and communication technologies adoption. To address these challenges, this paper proposes an innovative model-free resilience scheduling method based on state-adversarial deep reinforcement learning (DRL) for integrated demand response (IDR)-enabled IES. The proposed method designs an IDR program to explore the interaction ability of electricity-gas-heat flexible loads. Additionally, a state-adversarial Markov decision process (SA-MDP) model characterizes the energy scheduling problem of IES under cyber-attack. The state-adversarial soft actor-critic (SA-SAC) algorithm is proposed to mitigate the impact of cyber-attacks on the scheduling strategy. Simulation results demonstrate that our method is capable of adequately addressing the uncertainties resulting from RES and loads, mitigating the impact of cyber-attacks on the scheduling strategy, and ensuring a stable demand supply for various energy sources. Moreover, the proposed method demonstrates resilience against cyber-attacks. Compared to the original soft actor-critic (SAC) algorithm, it achieves a 10\% improvement in economic performance under cyber-attack scenarios.

Solving complex planning problems has been a long-standing challenge in computer science. Learning-based subgoal search methods have shown promise in tackling these problems, but they often suffer from a lack of completeness guarantees, meaning that they may fail to find a solution even if one exists. In this paper, we propose an efficient approach to augment a subgoal search method to achieve completeness in discrete action spaces. Specifically, we augment the high-level search with low-level actions to execute a multi-level (hybrid) search, which we call complete subgoal search. This solution achieves the best of both worlds: the practical efficiency of high-level search and the completeness of low-level search. We apply the proposed search method to a recently proposed subgoal search algorithm and evaluate the algorithm trained on offline data on complex planning problems. We demonstrate that our complete subgoal search not only guarantees completeness but can even improve performance in terms of search expansions for instances that the high-level could solve without low-level augmentations. Our approach makes it possible to apply subgoal-level planning for systems where completeness is a critical requirement.

One of the fundamental challenges associated with reinforcement learning (RL) is that collecting sufficient data can be both time-consuming and expensive. In this paper, we formalize a concept of time reversal symmetry in a Markov decision process (MDP), which builds upon the established structure of dynamically reversible Markov chains (DRMCs) and time-reversibility in classical physics. Specifically, we investigate the utility of this concept in reducing the sample complexity of reinforcement learning. We observe that utilizing the structure of time reversal in an MDP allows every environment transition experienced by an agent to be transformed into a feasible reverse-time transition, effectively doubling the number of experiences in the environment. To test the usefulness of this newly synthesized data, we develop a novel approach called time symmetric data augmentation (TSDA) and investigate its application in both proprioceptive and pixel-based state within the realm of off-policy, model-free RL. Empirical evaluations showcase how these synthetic transitions can enhance the sample efficiency of RL agents in time reversible scenarios without friction or contact. We also test this method in more realistic environments where these assumptions are not globally satisfied. We find that TSDA can significantly degrade sample efficiency and policy performance, but can also improve sample efficiency under the right conditions. Ultimately we conclude that time symmetry shows promise in enhancing the sample efficiency of reinforcement learning and provide guidance when the environment and reward structures are of an appropriate form for TSDA to be employed effectively.

Dynamic obstacle avoidance (DOA) is a fundamental challenge for any autonomous vehicle, independent of whether it operates in sea, air, or land. This paper proposes a two-step architecture for handling DOA tasks by combining supervised and reinforcement learning (RL). In the first step, we introduce a data-driven approach to estimate the collision risk of an obstacle using a recurrent neural network, which is trained in a supervised fashion and offers robustness to non-linear obstacle movements. In the second step, we include these collision risk estimates into the observation space of an RL agent to increase its situational awareness.~We illustrate the power of our two-step approach by training different RL agents in a challenging environment that requires to navigate amid multiple obstacles. The non-linear movements of obstacles are exemplarily modeled based on stochastic processes and periodic patterns, although our architecture is suitable for any obstacle dynamics. The experiments reveal that integrating our collision risk metrics into the observation space doubles the performance in terms of reward, which is equivalent to halving the number of collisions in the considered environment. Furthermore, we show that the architecture's performance improvement is independent of the applied RL algorithm.

This paper proposes a realistic modularized framework for controlling autonomous surface vehicles (ASVs) on inland waterways (IWs) based on deep reinforcement learning (DRL). The framework comprises two levels: a high-level local path planning (LPP) unit and a low-level path following (PF) unit, each consisting of a DRL agent. The LPP agent is responsible for planning a path under consideration of nearby vessels, traffic rules, and the geometry of the waterway. We thereby transfer a recently proposed spatial-temporal recurrent neural network architecture to continuous action spaces. The LPP agent improves operational safety in comparison to a state-of-the-art artificial potential field method by increasing the minimum distance to other vessels by 65% on average. The PF agent performs low-level actuator control while accounting for shallow water influences and the environmental forces winds, waves, and currents. Compared with a proportional-integral-derivative (PID) controller, the PF agent yields only 61% of the mean cross-track error while significantly reducing control effort in terms of the required absolute rudder angle. Lastly, both agents are jointly validated in simulation, employing the lower Elbe in northern Germany as an example case and using real automatic identification system (AIS) trajectories to model the behavior of other ships.

This review paper takes a comprehensive look at malicious attacks against FL, categorizing them from new perspectives on attack origins and targets, and providing insights into their methodology and impact. In this survey, we focus on threat models targeting the learning process of FL systems. Based on the source and target of the attack, we categorize existing threat models into four types, Data to Model (D2M), Model to Data (M2D), Model to Model (M2M) and composite attacks. For each attack type, we discuss the defense strategies proposed, highlighting their effectiveness, assumptions and potential areas for improvement. Defense strategies have evolved from using a singular metric to excluding malicious clients, to employing a multifaceted approach examining client models at various phases. In this survey paper, our research indicates that the to-learn data, the learning gradients, and the learned model at different stages all can be manipulated to initiate malicious attacks that range from undermining model performance, reconstructing private local data, and to inserting backdoors. We have also seen these threat are becoming more insidious. While earlier studies typically amplified malicious gradients, recent endeavors subtly alter the least significant weights in local models to bypass defense measures. This literature review provides a holistic understanding of the current FL threat landscape and highlights the importance of developing robust, efficient, and privacy-preserving defenses to ensure the safe and trusted adoption of FL in real-world applications.