Coverage conditions -- which assert that the data logging distribution adequately covers the state space -- play a fundamental role in determining the sample complexity of offline reinforcement learning. While such conditions might seem irrelevant to online reinforcement learning at first glance, we establish a new connection by showing -- somewhat surprisingly -- that the mere existence of a data distribution with good coverage can enable sample-efficient online RL. Concretely, we show that coverability -- that is, existence of a data distribution that satisfies a ubiquitous coverage condition called concentrability -- can be viewed as a structural property of the underlying MDP, and can be exploited by standard algorithms for sample-efficient exploration, even when the agent does not know said distribution. We complement this result by proving that several weaker notions of coverage, despite being sufficient for offline RL, are insufficient for online RL. We also show that existing complexity measures for online RL, including Bellman rank and Bellman-Eluder dimension, fail to optimally capture coverability, and propose a new complexity measure, the sequential extrapolation coefficient, to provide a unification.
Partial Observability -- where agents can only observe partial information about the true underlying state of the system -- is ubiquitous in real-world applications of Reinforcement Learning (RL). Theoretically, learning a near-optimal policy under partial observability is known to be hard in the worst case due to an exponential sample complexity lower bound. Recent work has identified several tractable subclasses that are learnable with polynomial samples, such as Partially Observable Markov Decision Processes (POMDPs) with certain revealing or decodability conditions. However, this line of research is still in its infancy, where (1) unified structural conditions enabling sample-efficient learning are lacking; (2) existing sample complexities for known tractable subclasses are far from sharp; and (3) fewer sample-efficient algorithms are available than in fully observable RL. This paper advances all three aspects above for Partially Observable RL in the general setting of Predictive State Representations (PSRs). First, we propose a natural and unified structural condition for PSRs called \emph{B-stability}. B-stable PSRs encompasses the vast majority of known tractable subclasses such as weakly revealing POMDPs, low-rank future-sufficient POMDPs, decodable POMDPs, and regular PSRs. Next, we show that any B-stable PSR can be learned with polynomial samples in relevant problem parameters. When instantiated in the aforementioned subclasses, our sample complexities improve substantially over the current best ones. Finally, our results are achieved by three algorithms simultaneously: Optimistic Maximum Likelihood Estimation, Estimation-to-Decisions, and Model-Based Optimistic Posterior Sampling. The latter two algorithms are new for sample-efficient learning of POMDPs/PSRs.
Finding unified complexity measures and algorithms for sample-efficient learning is a central topic of research in reinforcement learning (RL). The Decision-Estimation Coefficient (DEC) is recently proposed by Foster et al. (2021) as a necessary and sufficient complexity measure for sample-efficient no-regret RL. This paper makes progress towards a unified theory for RL with the DEC framework. First, we propose two new DEC-type complexity measures: Explorative DEC (EDEC), and Reward-Free DEC (RFDEC). We show that they are necessary and sufficient for sample-efficient PAC learning and reward-free learning, thereby extending the original DEC which only captures no-regret learning. Next, we design new unified sample-efficient algorithms for all three learning goals. Our algorithms instantiate variants of the Estimation-To-Decisions (E2D) meta-algorithm with a strong and general model estimation subroutine. Even in the no-regret setting, our algorithm E2D-TA improves upon the algorithms of Foster et al. (2021) which require either bounding a variant of the DEC which may be prohibitively large, or designing problem-specific estimation subroutines. As applications, we recover existing and obtain new sample-efficient learning results for a wide range of tractable RL problems using essentially a single algorithm. Finally, as a connection, we re-analyze two existing optimistic model-based algorithms based on Posterior Sampling or Maximum Likelihood Estimation, showing that they enjoy similar regret bounds as E2D-TA under similar structural conditions as the DEC.
A recent goal in the theory of deep learning is to identify how neural networks can escape the "lazy training," or Neural Tangent Kernel (NTK) regime, where the network is coupled with its first order Taylor expansion at initialization. While the NTK is minimax optimal for learning dense polynomials (Ghorbani et al, 2021), it cannot learn features, and hence has poor sample complexity for learning many classes of functions including sparse polynomials. Recent works have thus aimed to identify settings where gradient based algorithms provably generalize better than the NTK. One such example is the "QuadNTK" approach of Bai and Lee (2020), which analyzes the second-order term in the Taylor expansion. Bai and Lee (2020) show that the second-order term can learn sparse polynomials efficiently; however, it sacrifices the ability to learn general dense polynomials. In this paper, we analyze how gradient descent on a two-layer neural network can escape the NTK regime by utilizing a spectral characterization of the NTK (Montanari and Zhong, 2020) and building on the QuadNTK approach. We first expand upon the spectral analysis to identify "good" directions in parameter space in which we can move without harming generalization. Next, we show that a wide two-layer neural network can jointly use the NTK and QuadNTK to fit target functions consisting of a dense low-degree term and a sparse high-degree term -- something neither the NTK nor the QuadNTK can do on their own. Finally, we construct a regularizer which encourages our parameter vector to move in the "good" directions, and show that gradient descent on the regularized loss will converge to a global minimizer, which also has low test error. This yields an end to end convergence and generalization guarantee with provable sample complexity improvement over both the NTK and QuadNTK on their own.
This paper studies policy optimization algorithms for multi-agent reinforcement learning. We begin by proposing an algorithm framework for two-player zero-sum Markov Games in the full-information setting, where each iteration consists of a policy update step at each state using a certain matrix game algorithm, and a value update step with a certain learning rate. This framework unifies many existing and new policy optimization algorithms. We show that the state-wise average policy of this algorithm converges to an approximate Nash equilibrium (NE) of the game, as long as the matrix game algorithms achieve low weighted regret at each state, with respect to weights determined by the speed of the value updates. Next, we show that this framework instantiated with the Optimistic Follow-The-Regularized-Leader (OFTRL) algorithm at each state (and smooth value updates) can find an $\mathcal{\widetilde{O}}(T^{-5/6})$ approximate NE in $T$ iterations, which improves over the current best $\mathcal{\widetilde{O}}(T^{-1/2})$ rate of symmetric policy optimization type algorithms. We also extend this algorithm to multi-player general-sum Markov Games and show an $\mathcal{\widetilde{O}}(T^{-3/4})$ convergence rate to Coarse Correlated Equilibria (CCE). Finally, we provide a numerical example to verify our theory and investigate the importance of smooth value updates, and find that using "eager" value updates instead (equivalent to the independent natural policy gradient algorithm) may significantly slow down the convergence, even on a simple game with $H=2$ layers.
A conceptually appealing approach for learning Extensive-Form Games (EFGs) is to convert them to Normal-Form Games (NFGs). This approach enables us to directly translate state-of-the-art techniques and analyses in NFGs to learning EFGs, but typically suffers from computational intractability due to the exponential blow-up of the game size introduced by the conversion. In this paper, we address this problem in natural and important setups for the \emph{$\Phi$-Hedge} algorithm -- A generic algorithm capable of learning a large class of equilibria for NFGs. We show that $\Phi$-Hedge can be directly used to learn Nash Equilibria (zero-sum settings), Normal-Form Coarse Correlated Equilibria (NFCCE), and Extensive-Form Correlated Equilibria (EFCE) in EFGs. We prove that, in those settings, the \emph{$\Phi$-Hedge} algorithms are equivalent to standard Online Mirror Descent (OMD) algorithms for EFGs with suitable dilated regularizers, and run in polynomial time. This new connection further allows us to design and analyze a new class of OMD algorithms based on modifying its log-partition function. In particular, we design an improved algorithm with balancing techniques that achieves a sharp $\widetilde{\mathcal{O}}(\sqrt{XAT})$ EFCE-regret under bandit-feedback in an EFG with $X$ information sets, $A$ actions, and $T$ episodes. To our best knowledge, this is the first such rate and matches the information-theoretic lower bound.
Imperfect-Information Extensive-Form Games (IIEFGs) is a prevalent model for real-world games involving imperfect information and sequential plays. The Extensive-Form Correlated Equilibrium (EFCE) has been proposed as a natural solution concept for multi-player general-sum IIEFGs. However, existing algorithms for finding an EFCE require full feedback from the game, and it remains open how to efficiently learn the EFCE in the more challenging bandit feedback setting where the game can only be learned by observations from repeated playing. This paper presents the first sample-efficient algorithm for learning the EFCE from bandit feedback. We begin by proposing $K$-EFCE -- a more generalized definition that allows players to observe and deviate from the recommended actions for $K$ times. The $K$-EFCE includes the EFCE as a special case at $K=1$, and is an increasingly stricter notion of equilibrium as $K$ increases. We then design an uncoupled no-regret algorithm that finds an $\varepsilon$-approximate $K$-EFCE within $\widetilde{\mathcal{O}}(\max_{i}X_iA_i^{K}/\varepsilon^2)$ iterations in the full feedback setting, where $X_i$ and $A_i$ are the number of information sets and actions for the $i$-th player. Our algorithm works by minimizing a wide-range regret at each information set that takes into account all possible recommendation histories. Finally, we design a sample-based variant of our algorithm that learns an $\varepsilon$-approximate $K$-EFCE within $\widetilde{\mathcal{O}}(\max_{i}X_iA_i^{K+1}/\varepsilon^2)$ episodes of play in the bandit feedback setting. When specialized to $K=1$, this gives the first sample-efficient algorithm for learning EFCE from bandit feedback.
Few-shot abstractive summarization has become a challenging task in natural language generation. To support it, we designed a novel soft prompts architecture coupled with a prompt pre-training plus fine-tuning paradigm that is effective and tunes only extremely light parameters. The soft prompts include continuous input embeddings across an encoder and a decoder to fit the structure of the generation models. Importantly, a novel inner-prompt placed in the text is introduced to capture document-level information. The aim is to devote attention to understanding the document that better prompts the model to generate document-related content. The first step in the summarization procedure is to conduct prompt pre-training with self-supervised pseudo-data. This teaches the model basic summarizing capabilities. The model is then fine-tuned with few-shot examples. Experimental results on the CNN/DailyMail and XSum datasets show that our method, with only 0.1% of the parameters, outperforms full-model tuning where all model parameters are tuned. It also surpasses Prompt Tuning by a large margin and delivers competitive results against Prefix-Tuning with 3% of the parameters.
Non-autoregressive translation (NAT) models suffer from inferior translation quality due to removal of dependency on previous target tokens from inputs to the decoder. In this paper, we propose a novel and general approach to enhance the target dependency within the NAT decoder from two perspectives: decoder input and decoder self-attention. First, we transform the initial decoder input from the source language space to the target language space through a novel attentive transformation process. The transformation reassembles the decoder input based on target token embeddings and conditions the final output on the target-side information. Second, before NAT training, we introduce an effective forward-backward pre-training phase, implemented with different triangle attention masks. This pre-training phase enables the model to gradually learn bidirectional dependencies for the final NAT decoding process. Experimental results demonstrate that the proposed approaches consistently improve highly competitive NAT models on four WMT translation directions by up to 1.88 BLEU score, while overall maintaining inference latency comparable to other fully NAT models.