Low-Rank Markov Decision Processes (MDPs) have recently emerged as a promising framework within the domain of reinforcement learning (RL), as they allow for provably approximately correct (PAC) learning guarantees while also incorporating ML algorithms for representation learning. However, current methods for low-rank MDPs are limited in that they only consider finite action spaces, and give vacuous bounds as $|\mathcal{A}| \to \infty$, which greatly limits their applicability. In this work, we study the problem of extending such methods to settings with continuous actions, and explore multiple concrete approaches for performing this extension. As a case study, we consider the seminal FLAMBE algorithm (Agarwal et al., 2020), which is a reward-agnostic method for PAC RL with low-rank MDPs. We show that, without any modifications to the algorithm, we obtain a similar PAC bound when actions are allowed to be continuous. Specifically, when the model for transition functions satisfies a Hölder smoothness condition w.r.t. actions, and either the policy class has a uniformly bounded minimum density or the reward function is also Hölder smooth, we obtain a polynomial PAC bound that depends on the order of smoothness.

Probabilistic time series forecasting is a challenging problem due to the long sequences involved, the large number of samples needed for accurate probabilistic inference, and the need for real-time inference in many applications. These challenges necessitate methods that are not only accurate but computationally efficient. Unfortunately, most current state-of-the-art methods for time series forecasting are based on Transformers, which scale poorly due to quadratic complexity in sequence length, and are therefore needlessly computationally inefficient. Moreover, with a few exceptions, these methods have only been evaluated for non-probabilistic point estimation. In this work, we address these two shortcomings. For the first, we introduce VQ-TR, which maps large sequences to a discrete set of latent representations as part of the Attention module. This not only allows us to attend over larger context windows with linear complexity in sequence length but also allows for effective regularization to avoid overfitting. For the second, we provide what is to the best of our knowledge the first systematic comparison of modern Transformer-based time series forecasting methods for probabilistic forecasting. In this comparison, we find that VQ-TR performs better or comparably to all other methods while being computationally efficient.

We study the evaluation of a policy under best- and worst-case perturbations to a Markov decision process (MDP), using transition observations from the original MDP, whether they are generated under the same or a different policy. This is an important problem when there is the possibility of a shift between historical and future environments, \emph{e.g.} due to unmeasured confounding, distributional shift, or an adversarial environment. We propose a perturbation model that allows changes in the transition kernel densities up to a given multiplicative factor or its reciprocal, extending the classic marginal sensitivity model (MSM) for single time-step decision-making to infinite-horizon RL. We characterize the sharp bounds on policy value under this model -- \emph{i.e.}, the tightest possible bounds based on transition observations from the original MDP -- and we study the estimation of these bounds from such transition observations. We develop an estimator with several important guarantees: it is semiparametrically efficient, and remains so even when certain necessary nuisance functions, such as worst-case Q-functions, are estimated at slow, nonparametric rates. Our estimator is also asymptotically normal, enabling straightforward statistical inference using Wald confidence intervals. Moreover, when certain nuisances are estimated inconsistently, the estimator still provides valid, albeit possibly not sharp, bounds on the policy value. We validate these properties in numerical simulations. The combination of accounting for environment shifts from train to test (robustness), being insensitive to nuisance-function estimation (orthogonality), and addressing the challenge of learning from finite samples (inference) together leads to credible and reliable policy evaluation.

Educators should provide access to all students with inclusive and equitable computer science (CS) and computational thinking (CT) learning outcomes. Yet access to CS and CT is not always available for students with disabilities. This qualitative cross-case study examined the barriers and strengths three exemplar teachers faced, explored the supports and resources they provided, and presented how these teachers defined successful inclusive CS learning outcomes for their students. The data set included analysis of the semistructured teachers' interviews and teaching materials. The results included seven successful strategies the teachers identified: 1) using physical computing, 2) pair programming, 3)connecting CS and Individual Education Plans (IEP), 4)applying hands-on activities, 5) CT integration, 6) using CS vocabulary, and 7)open-ended pedagogy. Three resources and supports the teacher provided emerged from the data set as follows: 1) accessible instructional materials, 2) projects with multiple entry points, and 3) essential scaffolding supports. Whereas four barriers teachers faced; 1. subject matter, 2) accessible tools, 3) students receiving support, 3) the role of CS in instruction, and 4) the role of time served as additional findings. The results suggested that school-based practitioners, including administrators, can overcome the barriers and promote successful strategies that lead to asset-based CS-inclusion in the classroom.

Safety is a crucial necessity in many applications of reinforcement learning (RL), whether robotic, automotive, or medical. Many existing approaches to safe RL rely on receiving numeric safety feedback, but in many cases this feedback can only take binary values; that is, whether an action in a given state is safe or unsafe. This is particularly true when feedback comes from human experts. We therefore consider the problem of provable safe RL when given access to an offline oracle providing binary feedback on the safety of state, action pairs. We provide a novel meta algorithm, SABRE, which can be applied to any MDP setting given access to a blackbox PAC RL algorithm for that setting. SABRE applies concepts from active learning to reinforcement learning to provably control the number of queries to the safety oracle. SABRE works by iteratively exploring the state space to find regions where the agent is currently uncertain about safety. Our main theoretical results shows that, under appropriate technical assumptions, SABRE never takes unsafe actions during training, and is guaranteed to return a near-optimal safe policy with high probability. We provide a discussion of how our meta-algorithm may be applied to various settings studied in both theoretical and empirical frameworks.

In a variety of applications, including nonparametric instrumental variable (NPIV) analysis, proximal causal inference under unmeasured confounding, and missing-not-at-random data with shadow variables, we are interested in inference on a continuous linear functional (e.g., average causal effects) of nuisance function (e.g., NPIV regression) defined by conditional moment restrictions. These nuisance functions are generally weakly identified, in that the conditional moment restrictions can be severely ill-posed as well as admit multiple solutions. This is sometimes resolved by imposing strong conditions that imply the function can be estimated at rates that make inference on the functional possible. In this paper, we study a novel condition for the functional to be strongly identified even when the nuisance function is not; that is, the functional is amenable to asymptotically-normal estimation at root-n-rates. The condition implies the existence of debiasing nuisance functions, and we propose penalized minimax estimators for both the primary and debiasing nuisance functions. The proposed nuisance estimators can accommodate flexible function classes, and importantly they can converge to fixed limits determined by the penalization regardless of the identifiability of the nuisances. We use the penalized nuisance estimators to form a debiased estimator for the functional of interest and prove its asymptotic normality under generic high-level conditions, which provide for asymptotically valid confidence intervals. We also illustrate our method in a novel partially linear proximal causal inference problem and a partially linear instrumental variable regression problem.

In this paper, we study nonparametric estimation of instrumental variable (IV) regressions. Recently, many flexible machine learning methods have been developed for instrumental variable estimation. However, these methods have at least one of the following limitations: (1) restricting the IV regression to be uniquely identified; (2) only obtaining estimation error rates in terms weak metrics (e.g., projected norm) rather than strong metrics (e.g., L_2 norm); or (3) imposing the so-called closedness condition that requires a certain conditional expectation operator to be sufficiently smooth. In this paper, we present the first method and analysis that can avoid all three limitations, while still permitting general function approximation. Specifically, we propose a new penalized minimax estimator that can converge to a fixed IV solution even when there are multiple solutions, and we derive a strong L_2 error rate for our estimator under lax conditions. Notably, this guarantee only needs a widely-used source condition and realizability assumptions, but not the so-called closedness condition. We argue that the source condition and the closedness condition are inherently conflicting, so relaxing the latter significantly improves upon the existing literature that requires both conditions. Our estimator can achieve this improvement because it builds on a novel formulation of the IV estimation problem as a constrained optimization problem.

We study off-policy evaluation (OPE) for partially observable MDPs (POMDPs) with general function approximation. Existing methods such as sequential importance sampling estimators and fitted-Q evaluation suffer from the curse of horizon in POMDPs. To circumvent this problem, we develop a novel model-free OPE method by introducing future-dependent value functions that take future proxies as inputs. Future-dependent value functions play similar roles as classical value functions in fully-observable MDPs. We derive a new off-policy Bellman equation for future-dependent value functions as conditional moment equations that use history proxies as instrumental variables. We further propose a minimax learning method to learn future-dependent value functions using the new Bellman equation. We obtain the PAC result, which implies our OPE estimator is close to the true policy value as long as futures and histories contain sufficient information about latent states, and the Bellman completeness. Our code is available at https://github.com/aiueola/neurips2023-future-dependent-ope

Off-policy evaluation (OPE) in reinforcement learning is an important problem in settings where experimentation is limited, such as healthcare. But, in these very same settings, observed actions are often confounded by unobserved variables making OPE even more difficult. We study an OPE problem in an infinite-horizon, ergodic Markov decision process with unobserved confounders, where states and actions can act as proxies for the unobserved confounders. We show how, given only a latent variable model for states and actions, policy value can be identified from off-policy data. Our method involves two stages. In the first, we show how to use proxies to estimate stationary distribution ratios, extending recent work on breaking the curse of horizon to the confounded setting. In the second, we show optimal balancing can be combined with such learned ratios to obtain policy value while avoiding direct modeling of reward functions. We establish theoretical guarantees of consistency and benchmark our method empirically.

Recent work on policy learning from observational data has highlighted the importance of efficient policy evaluation and has proposed reductions to weighted (cost-sensitive) classification. But, efficient policy evaluation need not yield efficient estimation of policy parameters. We consider the estimation problem given by a weighted surrogate-loss classification with any score function, either direct, inverse-propensity-weighted, or doubly robust. We show that, under a correct specification assumption, the weighted classification formulation need not be efficient for policy parameters. We draw a contrast to actual (possibly weighted) binary classification, where correct specification implies a parametric model, while for policy learning it only implies a semi-parametric model. In light of this, we instead propose an estimation approach based on generalized method of moments, which is efficient for the policy parameters. We propose a particular method based on recent developments on solving moment problems using neural networks and demonstrate the efficiency and regret benefits of this method empirically.

Evaluating novel contextual bandit policies using logged data is crucial in applications where exploration is costly, such as medicine. But it usually relies on the assumption of no unobserved confounders, which is bound to fail in practice. We study the question of policy evaluation when we instead have proxies for the latent confounders and develop an importance weighting method that avoids fitting a latent outcome regression model. Surprisingly, we show that there exist no single set of weights that give unbiased evaluation regardless of outcome model, unlike the case with no unobserved confounders where density ratios are sufficient. Instead, we propose an adversarial objective and weights that minimize it, ensuring sufficient balance in the latent confounders regardless of outcome model. We develop theory characterizing the consistency of our method and tractable algorithms for it. Empirical results validate the power of our method when confounders are latent.

Instrumental variable analysis is a powerful tool for estimating causal effects when randomization or full control of confounders is not possible. The application of standard methods such as 2SLS, GMM, and more recent variants are significantly impeded when the causal effects are complex, the instruments are high-dimensional, and/or the treatment is high-dimensional. In this paper, we propose the DeepGMM algorithm to overcome this. Our algorithm is based on a new variational reformulation of GMM with optimal inverse-covariance weighting that allows us to efficiently control very many moment conditions. We further develop practical techniques for optimization and model selection that make it particularly successful in practice. Our algorithm is also computationally tractable and can handle large-scale datasets. Numerical results show our algorithm matches the performance of the best tuned methods in standard settings and continues to work in high-dimensional settings where even recent methods break.

We propose to decompose instruction execution to goal prediction and action generation.We design a model that maps raw visual observations to goals using LINGUNET, a language-conditioned image generation network, and then generates the actions required to complete them.Our model is trained from demonstration only without external resources.To evaluate our approach, we introduce two benchmarks for instruction following: LANI, a navigation task; and CHAI, where an agent executes household instructions.Our evaluation demonstrates the advantages of our model decomposition, and illustrates the challenges posed by our new benchmarks.

There is a class of complex problems where solutions must satisfy multiple subjective criteria, while meeting specific quantifiable constraints. Route planning for leisurely travel is an example of a problem in this class. Constraints including total available time, transit times, and one's budget and subjective interests determine whether a potential solution is acceptable to a prospective traveler. In this paper we present a route planning (routing) interface that metaphorically leverages various elastic properties of a rubber band to allow for playful interaction with the relevant constraints. Each of these properties — attenuation, tension, and color — were integrated into an experimental system and then investigated in a series of task-based evaluations. Our research shows this playful interaction enables potential travelers to explore the solution space in order to find a route that meets, not only the easily quantifiable constraints, but also their own subjective preferences.

There is a class of complex problems where solutions must satisfy multiple subjective criteria, while meeting specific quantifiable constraints. Route planning for leisurely travel is an example of a problem in this class, where constraints including total available time, transit times, and budget constraints determine whether a potential solution is acceptable to the prospective traveller. In this paper we present an interface that leverages, metaphorically, the elastic properties of a rubber band to allow playful interaction with relevant constraints. The resulting touch-based human computer interface enables the traveller to explore the solution space in the sense that constraints can be played with in order to find a route that meets the traveller's subjective preferences. Formal step-by-step evaluation with nine subjects confirms that leveraging the rubber band metaphor is useful in constraint satisfaction and that using the interface is intuitive since it leverages real-world experiences.

Clinical interviews were held with students in senior level electrical engineering and methods of teaching secondary mathematics classes. These students had all completed a 4-semester calculus sequence including differential equations. Engineering students had then taken advanced classes applying mathematical ideas in real-world contexts while mathematics education students had taken a similar amount of coursework in advanced mathematics courses. Students were interviewed to determine how their conceptual understanding of function and accumulation (integration) had grown or regressed during work after calculus. Back-transfer, where later learning improves conceptual understanding of earlier material, was observed.

We are developing online interactive materials that allow students to adapt their text and homework to their own needs and interests. For example, students may specify topic areas of interest and then selected mathematics homework word problems are adapted to those areas. Using proper data management, their work can be hand-graded with feedback for the students and allowances for students to correct and resubmit their work after initial feedback in less time than is taken in grading traditional assignments. Such tools can support individualizing instruction even in large lecture classes.

While classes in IT skills are endemic, high school students in the UK rarely experience computer science. We present a pilot of a scheme that aims to go some way towards addressing this. Specifically, computing clubs were run on high school premises by high school teachers using material prepared by the University of Leeds School of Computing and supported by volunteer undergraduate mentors. Feedback suggests that the clubs were highly successful in their objectives of broadening understanding of the idea of a computer and introducing the concept of a computer program. School students, their teachers and the undergraduate volunteers all report an enjoyable, purposeful experience.