Reinforcement learning (RL) is one of the most important paradigms in machine learning. What makes RL different from other paradigms is that it models the long-term effects in decision-making problems. For instance, in a finite-horizon Markov decision process (MDP), which is one of the most fundamental models for RL, an agent interacts with the environment for a total of H steps and receives a sequence of H random reward values, along with stochastic state transitions, as feedback. The goal of the agent is to find a policy to maximize the expected sum of these rewards values instead of any single one of them. Since decisions made at early stages could significantly impact the future, the agent must take possible future transitions into consideration when choosing the policy. On the other hand, when H 1, RL reduces to the contextual bandits problem in which it suffices to act myopically to achieve optimality. Due to the important role of the horizon length in RL, Jiang and Agarwal [JA18] propose to study how the sample complexity of RL depends on the horizon length. More formally, let us consider the episodic RL setting, where the horizon length is H and the underlying MDP has unknown and time invariant transition probabilities and rewards.

We address the issue of safety in reinforcement learning. We pose the problem in a discounted infinite-horizon constrained Markov decision process framework. Existing results have shown that gradient-based methods are able to achieve an $\mathcal{O}(1/\sqrt{T})$ global convergence rate both for the optimality gap and the constraint violation. We exhibit a natural policy gradient-based algorithm that has a faster convergence rate $\mathcal{O}(\log(T)/T)$ for both the optimality gap and the constraint violation. When Slater's condition is satisfied and known a priori, zero constraint violation can be further guaranteed for a sufficiently large $T$ while maintaining the same convergence rate.

After the user provides a few examples of desired outcomes, MURAL automatically infers a reward function that takes into account these examples and the agent's uncertainty for each state. Although reinforcement learning has shown success in domains such as robotics, chip placement and playing video games, it is usually intractable in its most general form. In particular, deciding when and how to visit new states in the hopes of learning more about the environment can be challenging, especially when the reward signal is uninformative. These questions of reward specification and exploration are closely connected -- the more directed and "well shaped" a reward function is, the easier the problem of exploration becomes. The answer to the question of how to explore most effectively is likely to be closely informed by the particular choice of how we specify rewards.

Modelling dependencies between climate extremes is important for climate risk assessment, for instance when allocating emergency management funds. In statistics, multivariate extreme value theory is often used to model spatial extremes. However, most commonly used approaches require strong assumptions and are either too simplistic or over-parametrised. From a machine learning perspective, Generative Adversarial Networks (GANs) are a powerful tool to model dependencies in high-dimensional spaces. Yet in the standard setting, GANs do not well represent dependencies in the extremes. Here we combine GANs with extreme value theory (evtGAN) to model spatial dependencies in summer maxima of temperature and winter maxima in precipitation over a large part of western Europe. We use data from a stationary 2000-year climate model simulation to validate the approach and explore its sensitivity to small sample sizes. Our results show that evtGAN outperforms classical GANs and standard statistical approaches to model spatial extremes. Already with about 50 years of data, which corresponds to commonly available climate records, we obtain reasonably good performance. In general, dependencies between temperature extremes are better captured than dependencies between precipitation extremes due to the high spatial coherence in temperature fields. Our approach can be applied to other climate variables and can be used to emulate climate models when running very long simulations to determine dependencies in the extremes is deemed infeasible.

Linear structural causal models (SCMs) -- in which each observed variable is generated by a subset of the other observed variables as well as a subset of the exogenous sources -- are pervasive in causal inference and casual discovery. However, for the task of causal discovery, existing work almost exclusively focus on the submodel where each observed variable is associated with a distinct source with non-zero variance. This results in the restriction that no observed variable can deterministically depend on other observed variables or latent confounders. In this paper, we extend the results on structure learning by focusing on a subclass of linear SCMs which do not have this property, i.e., models in which observed variables can be causally affected by any subset of the sources, and are allowed to be a deterministic function of other observed variables or latent confounders. This allows for a more realistic modeling of influence or information propagation in systems. We focus on the task of causal discovery form observational data generated from a member of this subclass. We derive a set of necessary and sufficient conditions for unique identifiability of the causal structure. To the best of our knowledge, this is the first work that gives identifiability results for causal discovery under both latent confounding and deterministic relationships. Further, we propose an algorithm for recovering the underlying causal structure when the aforementioned conditions are satisfied. We validate our theoretical results both on synthetic and real datasets.

We study automated intrusion prevention using reinforcement learning. Following a novel approach, we formulate the problem of intrusion prevention as an (optimal) multiple stopping problem. This formulation gives us insight into the structure of optimal policies, which we show to have threshold properties. For most practical cases, it is not feasible to obtain an optimal defender policy using dynamic programming. We therefore develop a reinforcement learning approach to approximate an optimal policy. Our method for learning and validating policies includes two systems: a simulation system where defender policies are incrementally learned and an emulation system where statistics are produced that drive simulation runs and where learned policies are evaluated. We show that our approach can produce effective defender policies for a practical IT infrastructure of limited size. Inspection of the learned policies confirms that they exhibit threshold properties.

We consider the problem of generating valid knockoffs for knockoff filtering which is a statistical method that provides provable false discovery rate guarantees for any model selection procedure. To this end, we are motivated by recent advances in multivariate distribution-free goodness-of-fit tests namely, the rank energy (RE), that is derived using theoretical results characterizing the optimal maps in the Monge's Optimal Transport (OT) problem. However, direct use of use RE for learning generative models is not feasible because of its high computational and sample complexity, saturation under large support discrepancy between distributions, and non-differentiability in generative parameters. To alleviate these, we begin by proposing a variant of the RE, dubbed as soft rank energy (sRE), and its kernel variant called as soft rank maximum mean discrepancy (sRMMD) using entropic regularization of Monge's OT problem. We then use sRMMD to generate deep knockoffs and show via extensive evaluation that it is a novel and effective method to produce valid knockoffs, achieving comparable, or in some cases improved tradeoffs between detection power Vs false discoveries.

Multi-agent control is a central theme in the Cyber-Physical Systems (CPS). However, current control methods either receive non-Markovian states due to insufficient sensing and decentralized design, or suffer from poor convergence. This paper presents the Delayed Propagation Transformer (DePT), a new transformer-based model that specializes in the global modeling of CPS while taking into account the immutable constraints from the physical world. DePT induces a cone-shaped spatial-temporal attention prior, which injects the information propagation and aggregation principles and enables a global view. With physical constraint inductive bias baked into its design, our DePT is ready to plug and play for a broad class of multi-agent systems. The experimental results on one of the most challenging CPS -- network-scale traffic signal control system in the open world -- show that our model outperformed the state-of-the-art expert methods on synthetic and real-world datasets. Our codes are released at: https://github.com/VITA-Group/DePT.

From the previous article, we know that Bayesian Neural Network would treat the model weights and outputs as variables. Instead of finding a set of optimal estimates, we are fitting the probability distributions for them. But the problem is "How can we know what their distributions look like?" To answer this, you have to learn what prior, posterior, and Bayes' theorem are. In the following, we will use an example for illustration.

We develop methods for forming prediction sets in an online setting where the data generating distribution is allowed to vary over time in an unknown fashion. Our framework builds on ideas from conformal inference to provide a general wrapper that can be combined with any black box method that produces point predictions of the unseen label or estimated quantiles of its distribution. While previous conformal inference methods rely on the assumption that the data points are exchangeable, our adaptive approach provably achieves the desired coverage frequency over long-time intervals irrespective of the true data generating process. We accomplish this by modelling the distribution shift as a learning problem in a single parameter whose optimal value is varying over time and must be continuously re-estimated. We test our method, adaptive conformal inference, on two real world datasets and find that its predictions are robust to visible and significant distribution shifts.