We study the reinforcement learning for finite-horizon episodic Markov decision processes with adversarial reward and full information feedback, where the unknown transition probability function is a linear function of a given feature mapping. We propose an optimistic policy optimization algorithm with Bernstein bonus and show that it can achieve $\tilde{O}(dH\sqrt{T})$ regret, where $H$ is the length of the episode, $T$ is the number of interaction with the MDP and $d$ is the dimension of the feature mapping. Furthermore, we also prove a matching lower bound of $\tilde{\Omega}(dH\sqrt{T})$ up to logarithmic factors. To the best of our knowledge, this is the first computationally efficient, nearly minimax optimal algorithm for adversarial Markov decision processes with linear function approximation.

Policy gradient methods are widely used in solving two-player zero-sum games to achieve superhuman performance in practice. However, it remains elusive when they can provably find a near-optimal solution and how many samples and iterations are needed. The current paper studies natural extensions of Natural Policy Gradient algorithm for solving two-player zero-sum games where function approximation is used for generalization across states. We thoroughly characterize the algorithms' performance in terms of the number of samples, number of iterations, concentrability coefficients, and approximation error. To our knowledge, this is the first quantitative analysis of policy gradient methods with function approximation for two-player zero-sum Markov games.

Self-training is a standard approach to semi-supervised learning where the learner's own predictions on unlabeled data are used as supervision during training. In this paper, we reinterpret this label assignment process as an optimal transportation problem between examples and classes, wherein the cost of assigning an example to a class is mediated by the current predictions of the classifier. This formulation facilitates a practical annealing strategy for label assignment and allows for the inclusion of prior knowledge on class proportions via flexible upper bound constraints. The solutions to these assignment problems can be efficiently approximated using Sinkhorn iteration, thus enabling their use in the inner loop of standard stochastic optimization algorithms. We demonstrate the effectiveness of our algorithm on the CIFAR-10, CIFAR-100, and SVHN datasets in comparison with FixMatch, a state-of-the-art self-training algorithm. Additionally, we elucidate connections between our proposed algorithm and existing confidence thresholded self-training approaches in the context of homotopy methods in optimization. Our code is available at https://github.com/stanford-futuredata/sinkhorn-label-allocation.

Policy gradient gives rise to a rich class of reinforcement learning (RL) methods, for example the REINFORCE. Yet the best known sample complexity result for such methods to find an $\epsilon$-optimal policy is $\mathcal{O}(\epsilon^{-3})$, which is suboptimal. In this paper, we study the fundamental convergence properties and sample efficiency of first-order policy optimization method. We focus on a generalized variant of policy gradient method, which is able to maximize not only a cumulative sum of rewards but also a general utility function over a policy's long-term visiting distribution. By exploiting the problem's hidden convex nature and leveraging techniques from composition optimization, we propose a Stochastic Incremental Variance-Reduced Policy Gradient (SIVR-PG) approach that improves a sequence of policies to provably converge to the global optimal solution and finds an $\epsilon$-optimal policy using $\tilde{\mathcal{O}}(\epsilon^{-2})$ samples.

In performative prediction, predictions guide decision-making and hence can influence the distribution of future data. To date, work on performative prediction has focused on finding performatively stable models, which are the fixed points of repeated retraining. However, stable solutions can be far from optimal when evaluated in terms of the performative risk, the loss experienced by the decision maker when deploying a model. In this paper, we shift attention beyond performative stability and focus on optimizing the performative risk directly. We identify a natural set of properties of the loss function and model-induced distribution shift under which the performative risk is convex, a property which does not follow from convexity of the loss alone. Furthermore, we develop algorithms that leverage our structural assumptions to optimize the performative risk with better sample efficiency than generic methods for derivative-free convex optimization.

Epistemic uncertainty is the part of out-of-sample prediction error due to the lack of knowledge of the learner. Whereas previous work was focusing on model variance, we propose a principled approach for directly estimating epistemic uncertainty by learning to predict generalization error and subtracting an estimate of aleatoric uncertainty, i.e., intrinsic unpredictability. This estimator of epistemic uncertainty includes the effect of model bias and can be applied in non-stationary learning environments arising in active learning or reinforcement learning. In addition to demonstrating these properties of Direct Epistemic Uncertainty Prediction (DEUP), we illustrate its advantage against existing methods for uncertainty estimation on downstream tasks including sequential model optimization and reinforcement learning. We also evaluate the quality of uncertainty estimates from DEUP for probabilistic classification of images and for estimating uncertainty about synergistic drug combinations.

We study the function approximation aspect of distributionally robust optimization (DRO) based on probability metrics, such as the Wasserstein and the maximum mean discrepancy. Our analysis leverages the insight that existing DRO paradigms hinge on function majorants such as the Moreau-Yosida regularization (supremal convolution). Deviating from those, this paper instead proposes robust learning algorithms based on smooth function approximation and interpolation. Our methods are simple in forms and apply to general loss functions without knowing functional norms a priori. Furthermore, we analyze the DRO risk bound decomposition by leveraging smooth function approximators and the convergence rate for empirical kernel mean embedding.

Performing inference in graphs is a common task within several machine learning problems, e.g., image segmentation, community detection, among others. For a given undirected connected graph, we tackle the statistical problem of exactly recovering an unknown ground-truth binary labeling of the nodes from a single corrupted observation of each edge. Such problem can be formulated as a quadratic combinatorial optimization problem over the boolean hypercube, where it has been shown before that one can (with high probability and in polynomial time) exactly recover the ground-truth labeling of graphs that have an isoperimetric number that grows with respect to the number of nodes (e.g., complete graphs, regular expanders). In this work, we apply a powerful hierarchy of relaxations, known as the sum-of-squares (SoS) hierarchy, to the combinatorial problem. Motivated by empirical evidence on the improvement in exact recoverability, we center our attention on the degree-4 SoS relaxation and set out to understand the origin of such improvement from a graph theoretical perspective. We show that the solution of the dual of the relaxed problem is related to finding edge weights of the Johnson and Kneser graphs, where the weights fulfill the SoS constraints and intuitively allow the input graph to increase its algebraic connectivity. Finally, as byproduct of our analysis, we derive a novel Cheeger-type lower bound for the algebraic connectivity of graphs with signed edge weights.

Edge federated learning (FL) is an emerging machine learning paradigm that trains a global parametric model from distributed datasets via wireless communications. This paper proposes a unit-modulus over-the-air computation (UM-AirComp) framework to facilitate efficient edge federated learning, which simultaneously uploads local model parameters and updates global model parameters via analog beamforming. The proposed framework avoids sophisticated baseband signal processing, leading to low communication delays and implementation costs. A training loss bound of UM-AirComp is derived and two low-complexity algorithms, termed penalty alternating minimization (PAM) and accelerated gradient projection (AGP), are proposed to minimize the nonconvex nonsmooth loss bound. Simulation results show that the proposed UM-AirComp framework with PAM algorithm not only achieves a smaller mean square error of model parameters' estimation, training loss, and testing error, but also requires a significantly shorter runtime than that of other benchmark schemes. Moreover, the proposed UM-AirComp framework with AGP algorithm achieves satisfactory performance while reduces the computational complexity by orders of magnitude compared with existing optimization algorithms. Finally, we demonstrate the implementation of UM-AirComp in a vehicle-to-everything autonomous driving simulation platform. It is found that autonomous driving tasks are more sensitive to model parameter errors than other tasks since the former neural networks are more sophisticated containing sparser model parameters.

Natural-gradient descent on structured parameter spaces (e.g., low-rank covariances) is computationally challenging due to complicated inverse Fisher-matrix computations. We address this issue for optimization, inference, and search problems by using \emph{local-parameter coordinates}. Our method generalizes an existing evolutionary-strategy method, recovers Newton and Riemannian-gradient methods as special cases, and also yields new tractable natural-gradient algorithms for learning flexible covariance structures of Gaussian and Wishart-based distributions. We show results on a range of applications on deep learning, variational inference, and evolution strategies. Our work opens a new direction for scalable structured geometric methods via local parameterizations.