We provide a unified algorithmic framework for ensemble sampling in nonlinear contextual bandits and develop corresponding regret bounds for two most common nonlinear contextual bandit settings: Generalized Linear Ensemble Sampling (\texttt{GLM-ES}) for generalized linear bandits and Neural Ensemble Sampling (\texttt{Neural-ES}) for neural contextual bandits. Both methods maintain multiple estimators for the reward model parameters via maximum likelihood estimation on randomly perturbed data. We prove high-probability frequentist regret bounds of $\mathcal{O}(d^{3/2} \sqrt{T} + d^{9/2})$ for \texttt{GLM-ES} and $\mathcal{O}(\widetilde{d} \sqrt{T})$ for \texttt{Neural-ES}, where $d$ is the dimension of feature vectors, $\widetilde{d}$ is the effective dimension of a neural tangent kernel matrix, and $T$ is the number of rounds. These regret bounds match the state-of-the-art results of randomized exploration algorithms in nonlinear contextual bandit settings. In the theoretical analysis, we introduce techniques that address challenges specific to nonlinear models. Practically, we remove fixed-time horizon assumptions by developing anytime versions of our algorithms, suitable when $T$ is unknown. Finally, we empirically evaluate \texttt{GLM-ES}, \texttt{Neural-ES}, and their anytime variants, demonstrating strong performance. Overall, our results establish ensemble sampling as a provable and practical randomized exploration approach for nonlinear contextual bandits.

We derive a novel PAC-Bayesian generalization bound for reinforcement learning that explicitly accounts for Markov dependencies in the data, through the chain's mixing time. This contributes to overcoming challenges in obtaining generalization guarantees for reinforcement learning, where the sequential nature of data breaks the independence assumptions underlying classical bounds. Our bound provides non-vacuous certificates for modern off-policy algorithms like Soft Actor-Critic. We demonstrate the bound's practical utility through PB-SAC, a novel algorithm that optimizes the bound during training to guide exploration. Experiments across continuous control tasks show that our approach provides meaningful confidence certificates while maintaining competitive performance.

State aggregation aims to reduce the computational complexity of solving Markov Decision Processes (MDPs) while preserving the performance of the original system. A fundamental challenge lies in optimizing policies within the aggregated, or abstract, space such that the performance remains optimal in the ground MDP-a property referred to as {"}optimal policy equivalence {"}. This paper presents an abstraction framework based on the notion of homomorphism, in which two Markov chains are deemed homomorphic if their value functions exhibit a linear relationship. Within this theoretical framework, we establish a sufficient condition for the equivalence of optimal policy. We further examine scenarios where the sufficient condition is not met and derive an upper bound on the approximation error and a performance lower bound for the objective function under the ground MDP. We propose Homomorphic Policy Gradient (HPG), which guarantees optimal policy equivalence under sufficient conditions, and its extension, Error-Bounded HPG (EBHPG), which balances computational efficiency and the performance loss induced by aggregation. In the experiments, we validated the theoretical results and conducted comparative evaluations against seven algorithms.

The empirical success of multi-agent reinforcement learning (MARL) has motivated the search for more efficient and scalable algorithms for large scale multi-agent systems. However, existing state-of-the-art algorithms do not fully exploit inter-agent coupling information to develop MARL algorithms. In this paper, we propose a systematic approach to leverage structures in the inter-agent couplings for efficient model-free reinforcement learning. We model the cooperative MARL problem via a Bayesian network and characterize the subset of agents, termed as the value dependency set, whose information is required by each agent to estimate its local action value function exactly. Moreover, we propose a partially decentralized training decentralized execution (P-DTDE) paradigm based on the value dependency set. We theoretically establish that the total variance of our P-DTDE policy gradient estimator is less than the centralized training decentralized execution (CTDE) policy gradient estimator. We derive a multi-agent policy gradient theorem based on the P-DTDE scheme and develop a scalable actor-critic algorithm. We demonstrate the efficiency and scalability of the proposed algorithm on multi-warehouse resource allocation and multi-zone temperature control examples. For dense value dependency sets, we propose an approximation scheme based on truncation of the Bayesian network and empirically show that it achieves a faster convergence than the exact value dependence set for applications with a large number of agents.

Over the past couple of decades, many active learning acquisition functions have been proposed, leaving practitioners with an unclear choice of which to use. Bayesian Decision Theory (BDT) offers a universal principle to guide decision-making. In this work, we derive BDT for (Bayesian) active learning in the myopic framework, where we imagine we only have one more point to label. This derivation leads to effective algorithms such as Expected Error Reduction (EER), Expected Predictive Information Gain (EPIG), and other algorithms that appear in the literature. Furthermore, we show that BAIT (active learning based on V-optimal experimental design) can be derived from BDT and asymptotic approximations. A key challenge of such methods is the difficult scaling to large batch sizes, leading to either computational challenges (BatchBALD) or dramatic performance drops (top-$B$ selection). Here, using a particular formulation of the decision process, we derive Partial Batch Label Sampling (ParBaLS) for the EPIG algorithm. We show experimentally for several datasets that ParBaLS EPIG gives superior performance for a fixed budget and Bayesian Logistic Regression on Neural Embeddings. Our code is available at https://github.com/ADDAPT-ML/ParBaLS.

Inverse problems, where the goal is to recover an unknown signal from noisy or incomplete measurements, are central to applications in medical imaging, remote sensing, and computational biology. Diffusion models have recently emerged as powerful priors for solving such problems. However, existing methods either rely on projection-based techniques that enforce measurement consistency through heuristic updates, or they approximate the likelihood $p(\boldsymbol{y} \mid \boldsymbol{x})$, often resulting in artifacts and instability under complex or high-noise conditions. To address these limitations, we propose a novel framework called \emph{coupled data and measurement space diffusion posterior sampling} (C-DPS), which eliminates the need for constraint tuning or likelihood approximation. C-DPS introduces a forward stochastic process in the measurement space $\{\boldsymbol{y}_t\}$, evolving in parallel with the data-space diffusion $\{\boldsymbol{x}_t\}$, which enables the derivation of a closed-form posterior $p(\boldsymbol{x}_{t-1} \mid \boldsymbol{x}_t, \boldsymbol{y}_{t-1})$. This coupling allows for accurate and recursive sampling based on a well-defined posterior distribution. Empirical results demonstrate that C-DPS consistently outperforms existing baselines, both qualitatively and quantitatively, across multiple inverse problem benchmarks.

Real-world clinical problems are often characterized by multimodal data, usually associated with incomplete views and limited sample sizes in their cohorts, posing significant limitations for machine learning algorithms. In this work, we propose a Bayesian approach designed to efficiently handle these challenges while providing interpretable solutions. Our approach integrates (1) a generative formulation to capture cross-view relationships with a semi-supervised strategy, and (2) a discriminative task-oriented formulation to identify relevant information for specific downstream objectives. This dual generative-discriminative formulation offers both general understanding and task-specific insights; thus, it provides an automatic imputation of the missing views while enabling robust inference across different data sources. The potential of this approach becomes evident when applied to the multimodal clinical data, where our algorithm is able to capture and disentangle the complex interactions among biological, psychological, and sociodemographic modalities.

Transformer-based models for amortized probabilistic inference, such as neural processes, prior-fitted networks, and tabular foundation models, excel at single-pass marginal prediction. However, many real-world applications, from signal interpolation to multi-column tabular predictions, require coherent joint distributions that capture dependencies between predictions. While purely autoregressive architectures efficiently generate such distributions, they sacrifice the flexible set-conditioning that makes these models powerful for meta-learning. Conversely, the standard approach to obtain joint distributions from set-based models requires expensive re-encoding of the entire augmented conditioning set at each autoregressive step. We introduce a causal autoregressive buffer that preserves the advantages of both paradigms. Our approach decouples context encoding from updating the conditioning set. The model processes the context once and caches it. A dynamic buffer then captures target dependencies: as targets are incorporated, they enter the buffer and attend to both the cached context and previously buffered targets. This enables efficient batched autoregressive generation and one-pass joint log-likelihood evaluation. A unified training strategy allows seamless integration of set-based and autoregressive modes at minimal additional cost. Across synthetic functions, EEG signals, cognitive models, and tabular data, our method matches predictive accuracy of strong baselines while delivering up to 20 times faster joint sampling. Our approach combines the efficiency of autoregressive generative models with the representational power of set-based conditioning, making joint prediction practical for transformer-based probabilistic models.

The vulnerability of machine learning models to adversarial attacks remains a critical security challenge. Traditional defenses, such as adversarial training, typically robustify models by minimizing a worst-case loss. However, these deterministic approaches do not account for uncertainty in the adversary's attack. While stochastic defenses placing a probability distribution on the adversary exist, they often lack statistical rigor and fail to make explicit their underlying assumptions. To resolve these issues, we introduce a formal Bayesian framework that models adversarial uncertainty through a stochastic channel, articulating all probabilistic assumptions. This yields two robustification strategies: a proactive defense enacted during training, aligned with adversarial training, and a reactive defense enacted during operations, aligned with adversarial purification. Several previous defenses can be recovered as limiting cases of our model. We empirically validate our methodology, showcasing the benefits of explicitly modeling adversarial uncertainty.

Autonomous navigation in unfamiliar environments requires robots to simultaneously explore, localise, and plan under uncertainty, without relying on predefined maps or extensive training. We present a biologically inspired, Active Inference-based framework, Active Inference MAPping and Planning (AIMAPP). This model unifies mapping, localisation, and decision-making within a single generative model. Inspired by hippocampal navigation, it uses topological reasoning, place-cell encoding, and episodic memory to guide behaviour. The agent builds and updates a sparse topological map online, learns state transitions dynamically, and plans actions by minimising Expected Free Energy. This allows it to balance goal-directed and exploratory behaviours. We implemented a ROS-compatible navigation system that is sensor and robot-agnostic, capable of integrating with diverse hardware configurations. It operates in a fully self-supervised manner, is resilient to drift, and supports both exploration and goal-directed navigation without any pre-training. We demonstrate robust performance in large-scale real and simulated environments against state-of-the-art planning models, highlighting the system's adaptability to ambiguous observations, environmental changes, and sensor noise. The model offers a biologically inspired, modular solution to scalable, self-supervised navigation in unstructured settings. AIMAPP is available at https://github.com/decide-ugent/AIMAPP.