We consider an online learning problem where the learner interacts with a Markov decision process in a sequence of episodes, where the reward function is allowed to change between episodes in an adversarial manner and the learner only gets to observe the rewards associated with its actions. We allow the state space to be arbitrarily large, but we assume that all action-value functions can be represented as linear functions in terms of a known low-dimensional feature map, and that the learner has access to a simulator of the environment that allows generating trajectories from the true MDP dynamics. Our main contribution is developing a computationally efficient algorithm that we call MDP-LinExp3, and prove that its regret is bounded by $\widetilde{\mathcal{O}}\big(H^2 T^{2/3} (dK)^{1/3}\big)$, where $T$ is the number of episodes, $H$ is the number of steps in each episode, $K$ is the number of actions, and $d$ is the dimension of the feature map. We also show that the regret can be improved to $\widetilde{\mathcal{O}}\big(H^2 \sqrt{TdK}\big)$ under much stronger assumptions on the MDP dynamics. To our knowledge, MDP-LinExp3 is the first provably efficient algorithm for this problem setting.

Resolving the exploration-exploitation trade-off remains a fundamental problem in the design and implementation of reinforcement learning (RL) algorithms. In this paper, we focus on model-free RL using the epsilon-greedy exploration policy, which despite its simplicity, remains one of the most frequently used forms of exploration. However, a key limitation of this policy is the specification of $\varepsilon$. In this paper, we provide a novel Bayesian perspective of $\varepsilon$ as a measure of the uniformity of the Q-value function. We introduce a closed-form Bayesian model update based on Bayesian model combination (BMC), based on this new perspective, which allows us to adapt $\varepsilon$ using experiences from the environment in constant time with monotone convergence guarantees. We demonstrate that our proposed algorithm, $\varepsilon$-\texttt{BMC}, efficiently balances exploration and exploitation on different problems, performing comparably or outperforming the best tuned fixed annealing schedules and an alternative data-dependent $\varepsilon$ adaptation scheme proposed in the literature.

Non-uniform sampling arises when an experimenter does not have full control over the sampling characteristics of the process under investigation. Moreover, it is introduced intentionally in algorithms such as Bayesian optimization and compressive sensing. We argue that Stochastic Differential Equations (SDEs) are especially well-suited for characterizing second order moments of such time series. We introduce new initial estimates for the numerical optimization of the likelihood, based on incremental estimation and initialization from autoregressive models. Furthermore, we introduce model truncation as a purely data-driven method to reduce the order of the estimated model based on the SDE likelihood. We show the increased accuracy achieved with the new estimator in simulation experiments, covering all challenging circumstances that may be encountered in characterizing a non-uniformly sampled time series. Finally, we apply the new estimator to experimental rainfall variability data.

Determination of inspection and maintenance policies for minimizing long-term risks and costs in deteriorating engineering environments constitutes a complex optimization problem. Major computational challenges include the (i) curse of dimensionality, due to exponential scaling of state/action set cardinalities with the number of components; (ii) curse of history, related to exponentially growing decision-trees with the number of decision-steps; (iii) presence of state uncertainties, induced by inherent environment stochasticity and variability of inspection/monitoring measurements; (iv) presence of constraints, pertaining to stochastic long-term limitations, due to resource scarcity and other infeasible/undesirable system responses. In this work, these challenges are addressed within a joint framework of constrained Partially Observable Markov Decision Processes (POMDP) and multi-agent Deep Reinforcement Learning (DRL). POMDPs optimally tackle (ii)-(iii), combining stochastic dynamic programming with Bayesian inference principles. Multi-agent DRL addresses (i), through deep function parametrizations and decentralized control assumptions. Challenge (iv) is herein handled through proper state augmentation and Lagrangian relaxation, with emphasis on life-cycle risk-based constraints and budget limitations. The underlying algorithmic steps are provided, and the proposed framework is found to outperform well-established policy baselines and facilitate adept prescription of inspection and intervention actions, in cases where decisions must be made in the most resource- and risk-aware manner.

Bayesian Networks (BNs) are graphical probabilistic models that have proven popular in medical applications. While numerous medical BNs have been published, most are presented fait accompli without explanation of how the network structure was developed or justification of why it represents the correct structure for the given medical application. This means that the process of building medical BNs from experts is typically ad hoc and offers little opportunity for methodological improvement. This paper proposes generally applicable and reusable medical reasoning patterns to aid those developing medical BNs. The proposed method complements and extends the idiom-based approach introduced by Neil, Fenton, and Nielsen in 2000. We propose instances of their generic idioms that are specific to medical BNs. We refer to the proposed medical reasoning patterns as medical idioms. In addition, we extend the use of idioms to represent interventional and counterfactual reasoning. We believe that the proposed medical idioms are logical reasoning patterns that can be combined, reused and applied generically to help develop medical BNs. All proposed medical idioms have been illustrated using medical examples on coronary artery disease. The method has also been applied to other ongoing BNs being developed with medical experts. Finally, we show that applying the proposed medical idioms to published BN models results in models with a clearer structure.

Deep learning models are widely used for various industrial and scientific applications. Even though these models have achieved considerable success in recent years, there exists a lack of understanding of the rationale behind decisions made by such systems in the machine learning community. This problem of interpretability is further aggravated by the increasing complexity of such models. This paper utilizes concepts from machine learning, quantum computation and quantum field theory to demonstrate how a many valued quantum logic system naturally arises in a specific class of generative deep learning models called Convolutional Deep Belief Networks. It provides a robust theoretical framework for constructing deep learning models equipped with the interpretability of many valued quantum logic systems without compromising their computing efficiency.

Classification approaches based on the direct estimation and analysis of posterior probabilities will degrade if the original class priors begin to change. We prove that a unique (up to scale) solution is possible to recover the data likelihoods for a test example from its original class posteriors and dataset priors. Given the recovered likelihoods and a set of new priors, the posteriors can be re-computed using Bayes' Rule to reflect the influence of the new priors. The method is simple to compute and allows a dynamic update of the original posteriors.

Stationary stochastic processes (SPs) are a key component of many probabilistic models, such as those for off-the-grid spatio-temporal data. They enable the statistical symmetry of underlying physical phenomena to be leveraged, thereby aiding generalization. Prediction in such models can be viewed as a translation equivariant map from observed data sets to predictive SPs, emphasizing the intimate relationship between stationarity and equivariance. Building on this, we propose the Convolutional Neural Process (ConvNP), which endows Neural Processes (NPs) with translation equivariance and extends convolutional conditional NPs to allow for dependencies in the predictive distribution. The latter enables ConvNPs to be deployed in settings which require coherent samples, such as Thompson sampling or conditional image completion. Moreover, we propose a new maximum-likelihood objective to replace the standard ELBO objective in NPs, which conceptually simplifies the framework and empirically improves performance. We demonstrate the strong performance and generalization capabilities of ConvNPs on 1D regression, image completion, and various tasks with real-world spatio-temporal data.

Structured stochastic processes evolving in continuous time present a widely adopted framework to model phenomena occurring in nature and engineering. However, such models are often chosen to satisfy the Markov property to maintain tractability. One of the more popular of such memoryless models are Continuous Time Bayesian Networks (CTBNs). In this work, we lift its restriction to exponential survival times to arbitrary distributions. Current extensions achieve this via auxiliary states, which hinder tractability. To avoid that, we introduce a set of node-wise clocks to construct a collection of graph-coupled semi-Markov chains. We provide algorithms for parameter and structure inference, which make use of local dependencies and conduct experiments on synthetic data and a data-set generated through a benchmark tool for gene regulatory networks. In doing so, we point out advantages compared to current CTBN extensions.

We study the inverse reinforcement learning (IRL) problem under the \emph{transition dynamics mismatch} between the expert and the learner. In particular, we consider the Maximum Causal Entropy (MCE) IRL learner model and provide an upper bound on the learner's performance degradation based on the $\ell_1$-distance between the two transition dynamics of the expert and the learner. Then, by leveraging insights from the Robust RL literature, we propose a robust MCE IRL algorithm, which is a principled approach to help with this mismatch issue. Finally, we empirically demonstrate the stable performance of our algorithm compared to the standard MCE IRL algorithm under transition mismatches in finite MDP problems.