Temporal Difference

Reliable temporal-difference targets are central to off-policy actor-critic learning. Direct value improvement refines the next-state target with alternative actions, but the reliability of this refinement depends on how candidate actions are ranked, reviewed, and weighted. Noisy rankings may force premature candidate commitment, reusing selection scores may bias target valuation, and fixed enhancement weights may amplify weak evidence. To address these risks, we develop Conservative Adaptive Ranking and Screening (CARS), which retains an ordered candidate prefix within a preset budget and narrows it only when the observed boundary gap exceeds a disagreement-scaled uncertainty radius. Selector-Evaluator Value Assessment (SEVA) uses selector critics to order candidates and a separately parameterized evaluator critic to review the selected value, then caps the reviewed value at the selector reference. Dynamic Adaptive Risk-aware Enhancement (DARE) then regulates each residual correction using candidate reliability, the gap between selector and evaluator signals, and a finite stage factor. Together, CARS, SEVA, and DARE form CARE-VI, an evidence-regulated target construction framework that preserves the backbone interfaces for critic regression and actor updates. The analysis bounds the CARS boundary error, the SEVA selected-value overestimation, and the one-sided deviation of the DARE residual displacement from its population counterpart, and establishes fixed-policy recovery after the finite-stage perturbation ends. Experiments with SAC, TD3, and TD7 on four MuJoCo tasks show that CARE-VI achieves the highest mean return in all twelve settings. Grouped ablations and scalar diagnostics support the roles of the three components in improving target reliability.

Quantile temporal-difference learning (QTD) is an effective method for learning return distributions through quantile approximation, yet its finite-time behavior remains poorly understood. Its update is nonlinear and nonsmooth, and the stability needed for a sharp convergence rate holds only near the target. We establish a global high-probability last-iterate guarantee for synchronous tabular QTD under general positive, nonincreasing step-size sequences and arbitrary initialization in the natural parameter range. For polynomially decaying step sizes with exponent $a\in(0,1)$, the last iterate converges to the target at rate $T^{-a/2}$ in the infinity norm, up to logarithmic and lower-order terms. A suitably tuned harmonic schedule recovers the $T^{-1/2}$ statistical rate up to logarithmic factors. For the $m$-quantile representation, its $\infty$-Wasserstein error scales as $\sqrt{m/T}$ up to logarithmic factors, matching the leading polynomial dependence on the quantile resolution and sample size of the corresponding model-based estimator. The proof uses a two-stage global-to-local argument. From arbitrary initialization, Bellman contraction and CDF monotonicity first bring the iterate close to the target, after which, a novel variance--drift matching argument sharpens the control of accumulated noise and local contraction reduces the remaining errors, yielding the sharp rate. Simulations verify the predicted polynomial decay and assess the finite-time entrance bound.

In this paper, we study how to perform statistical inference for quantile temporal difference learning (QTD) in distributional reinforcement learning. Assuming access to a generative model, we first establish functional central limit theorems for both synchronous and asynchronous QTD, which show that the averaged iterates of QTD converge weakly to a rescaled Brownian motion. We next provide online inference methods. Based on random scaling, the inference procedure constructs an asymptotically pivotal statistic for inference by using the information along the whole QTD path. Meanwhile, the proposed statistic can be computed online without storing the entire trajectory of QTD iterates. This substantially reduces the memory requirement and enables efficient statistical inference in distributional reinforcement learning.

Constant-stepsize temporal-difference (TD) learning is attractive for policy evaluation, but inference from a single Markov trajectory must account for serial dependence and a stepsize-dependent stationary target. For fixed-stepsize linear TD, we establish a functional central limit theorem whose covariance retains the multiplicative component induced by the random TD matrix and the stationary iterate error. We then derive a joint functional limit for parallel Richardson--Romberg (RR) recursions driven by the same trajectory. A Brownian-bridge self-normalizer yields asymptotically pivotal confidence regions for prespecified state-value contrasts without estimating the long-run covariance or selecting a bandwidth or batch length. For such a contrast, the procedure admits a one-pass implementation whose memory does not grow with the trajectory length. At a fixed stepsize, the inferential center is the RR stationary target. We also study horizon-indexed designs in which the stepsize remains constant within each run and decreases across longer horizons. Under an explicit RR-dependent rate window, the residual RR target shift, multiplicative remainder, and initialization effect are negligible at the root-$n$ scale, yielding inference for the projected Bellman solution. Experiments on FrozenLake and Garnet illustrate stationary-target coverage, RR target correction, and the finite-sample behavior of the horizon-indexed design.

Training large language model agents in long-horizon environments requires assigning credit from sparse terminal outcomes to individual actions. Existing critic-free methods propagate trajectory-level rewards uniformly across steps, while recent approaches construct step-level groups by matching repeated states and compare actions within each group. The former cannot distinguish useful actions in failed trajectories from ineffective actions in successful ones. The latter rely on step credit derived directly from individual trajectory outcomes and fixed-weight fusion with episode-level credit. We propose Gated-BEPO, which derives step-level credit from empirical rollout graphs. For each rollout group, Gated-BEPO constructs an empirical graph and estimates node values through a mean-backup Bellman fixed point that reflects the empirical action distribution of the current policy. We then accumulate these temporal-difference residuals along each sampled trajectory using generalized advantage estimation, yielding step-level Bellman advantages that capture both immediate and downstream effects. To adaptively fuse episode- and step-level credit, a confidence gate incorporates Bellman credit only at states with multiple observed successors and otherwise uses episode-level credit. Experiments on WebShop, ALFWorld, and visual Sokoban show consistent improvements across language and vision-language models, while diagnostic ablations support the effectiveness of Bellman fixed-point value estimation and show that step-level credit should be incorporated selectively rather than uniformly into the final advantage.

Video-to-audio (V2A) generation extends image-to-audio generation (I2A) by introducing consecutive frames that provide essential temporal cues for audio synthesis. However, existing conditional diffusion-based V2A methods typically enhance visual conditioning with additional audio-visual supervision, acoustic structure prediction, or reasoning from large multimodal models, requiring extra networks or strong inductive biases. Inspired by recent advances in visual representation learning, we introduce TD-V2A, which leverages temporal differences (TD) as the key representation that distinguishes V2A from I2A, enriching visual conditioning with minimal architectural modification. We first investigate TD at both the frame and feature levels to identify the most effective representation level at which TD complements visual representations. Based on these findings, we develop a hierarchically continual learning strategy and an annealed temporal differences guidance method to progressively learn and exploit TD information during diffusion training and sampling process, respectively. Extensive experiments on benchmark datasets demonstrate that effectively exploiting TD through our proposed framework significantly improves end-to-end V2A generation quality, even outperforming dedicated V2A representations such as contrastive audio-visual pretraining.

Deep Q-Networks (DQNs) learn value functions through bootstrapped temporal-difference updates, where future returns are approximated using a greedy maximization over next-state action values. While effective, this aggregation rule is inherently sensitive to estimation noise: when Q-values are uncertain, the maximization operator deterministically favors the largest estimate, regardless of its reliability, leading to amplified errors through bootstrapping. In this work, we propose the \textbf{S}uccessor Rollout \textbf{A}ggregation \textbf{D}eep \textbf{Q}-Network (SADQ), a simple modification to Q-learning that regularizes how the TD target is formed. SADQ uses one-step rollout predictions from a learned dynamics model to guide the comparison among candidate next-state actions, introducing additional structure into the aggregation step without altering the underlying learning framework. The resulting mixed Bellman update attenuates unreliable maxima while preserving the standard fixed point under diminishing model error. We provide theoretical analysis showing that SADQ reduces bootstrap-induced overestimation in a pointwise manner. Empirically, SADQ consistently improves training stability across classical control tasks, real-world vector-based environments, and Atari benchmarks when compared to strong DQN variants.

In this paper, we present a generalized temporal-difference (TD) reinforcement learning framework based on the theory of conditional expectations. The value and action-value (Q-value) functions are treated as uncertain quantities, and their estimation is formulated as a stochastic inference problem. Unlike classical Kalman-based temporal-difference learning, which relies on linear-Gaussian assumptions, the proposed formulation is derived directly from the conditional expectation framework and naturally extends to nonlinear models and non-Gaussian probability distributions. The proposed method recursively estimates not only the conditional expectation of the value function but also its second probabilistic moment, thereby quantifying the uncertainty associated with the learned value function throughout the learning process. To obtain a computationally tractable algorithm, the stochastic problem is discretized using either polynomial chaos expansions or ensemble-based approximations, providing efficient representations of the underlying random variables. The proposed framework is demonstrated on two optimal control problems: a linear mass--spring--damper system and a nonlinear heat conduction problem in a closed cavity. The numerical examples illustrate the capability of the proposed method to accurately estimate both the value function and its associated uncertainty, while extending classical Kalman-based temporal-difference learning to a broader class of stochastic systems.

End-to-end deep reinforcement learning (DRL) for zero-shot object-goal visual navigation remains challenged by the sim-to-real gap, particularly variations in object appearance and restricted camera field-of-view (FoV). This letter proposes a Temporal Difference-Relational Network (T-DRN) for robust zero-shot sim-to-real transfer. T-DRN combines a Siamese difference-based feature extractor, which computes relational difference between the target and observed objects to produce domain-independent representations, with a dual-frame temporal buffer that preserves short-term object continuity under narrow FoV. Extensive experiments in AI2-THOR demonstrate that T-DRN improves zero-shot generalization in terms of success rates over strong baselines. Furthermore, T-DRN is systematically validated on a physical wheeled robot, demonstrating robust performance under real sensing and actuation constraints and supporting the feasibility of direct sim-to-real transfer.

In this paper, we study Reinforcement Learning in Parametrized Action Markov Decision Processes (PAMDP), where each decision consists of a symbolic action and numerical parameters. In such settings Reinforcement Learning algorithms typically determine parameters with one-shot estimators, which makes their training sample inefficient. Though in most PAMDP environments explicit but incomplete knowledge (e.g., rules, safety constraints, or expert heuristics) is available, it is rarely directly used to increase the sample-efficiency of training Reinforcement Learning agents. We step into this gap and propose our novel Neuro-Symbolic Knowledge- and Gradient-Guided Reinforcement Learning (KGRL) algorithm. KGRL uses domain knowledge in a Datalog knowledge base to derive the set of applicable actions and feasible parameters for a given state. This allows it to prune non-applicable actions from the decision-space and constrain the parameter spaces of the remaining actions. We then use a gradient-based parameter refinement loop to estimate the optimal parameters during training and deployment of the agent. By recording activated rules along the trajectory, KGRL additionally provides local procedural explanations on the pruning of actions and constraining of parameters. Overall, KGRL guides the agent's exploration and deployment toward feasible and constraint-aware decisions, while increasing sample efficiency during training. KGRL outperforms state-of-the-art RL baselines for PAMDPs in both, sample efficiency and episodic return.

This work delivers two key contributions: one to efficient feature selection in reinforcement learning (RL), the other to the theory of non-monotone inclusions. On the RL side, the estimation bias inherent in conventional regularization schemes is addressed by augmenting classical least-squares temporal-difference (LSTD) policy evaluation with the sparsity-inducing, non-convex projected minimax concave (PMC) penalty. Because the PMC penalty is weakly convex, the resulting fixed-point problem is no longer monotone; instead, it falls under a broader class of non-monotone inclusions involving the sum of a monotone Lipschitz operator and a hypomonotone operator. On the theory side, novel convergence conditions are developed for the forward-reflected-backward splitting (FRBS) method applied to this broader class of non-monotone inclusion problems. Under mild conditions, Lyapunov stability and the existence of a limit point of the sequence of FRBS iterates are established; alternatively, under the weak Minty variational inequality assumption, exact convergence is guaranteed. Numerical tests on benchmark datasets show that the proposed FRBS iterates, applied to the non-convexly regularized LSTD problem, substantially outperform state-of-the-art feature-selection methods, especially when many noisy features are present.

Reinforcement learning in large or sparse-reward environments suffers from slow temporal-difference reward propagation, as value information spreads only locally across the state space. We propose Mesh-RL, a spatial domain-decomposition framework inspired by the finite element method and domain decomposition theory, which partitions the environment into overlapping subgrids and enforces boundary-consistent temporal-difference updates. Such an approach enables localized learning while ensuring globally coherent value propagation. Unlike hierarchical or model-based approaches, Mesh-RL accelerates long-range credit assignment without modifying the reward function, Bellman operator, or introducing explicit planning mechanisms. We evaluate Mesh-RL on hazard-dense grid-world environments with varying geometries and mesh resolutions. Across Q-learning, SARSA, and Dyna-Q, Mesh-RL consistently improves convergence speed, cumulative reward, and learning stability. Higher mesh resolutions sustain exploration, prevent premature convergence, and substantially accelerate value propagation to distant states. While Dyna-Q already benefits from internal planning, it still achieves additional gains under structured decomposition. Overall, Mesh-RL introduces a principled spatial domain-decomposition mechanism for accelerating temporal-difference learning. Our framework bridges finite element method-inspired boundary-consistency techniques from scientific computing with reinforcement learning to improve sample efficiency in sparse-reward environments. We will release source code of the study.

We analyze the variance of temporal difference (TD) learning using the phased setting with tabular representation, and show that one of the mechanisms behind its ability to reduce variance is by effectively aggregating over a larger number of independent trajectories. Based on this insight, we demonstrate that (1) the variance of TD is asymptotically bounded from above by Monte Carlo (MC) estimators, and (2) shorter horizon updates incurs less variance for a fixed number of samples. Beyond TD, we show that Direct Advantage Estimation (DAE), a method for estimating the advantage function, can be seen as a type of regression-adjusted control variate, which achieves a tighter bound on the variance compared to TD in the large-sample limit. Finally, we numerically illustrate the behaviors of these estimators with carefully designed environments.

Temporal difference (TD) learning with linear function approximation is a core method for policy evaluation. Its classical continuous-time description is an ordinary differential equation (ODE), which captures the asymptotic mean dynamics but neglects stochastic fluctuations determining the error floor. We introduce a stochastic differential equation (SDE) approximation for linear TD(0) under Markovian noise. The resulting model distinguishes the contraction dynamics governed by the projected Bellman operator from the influence of Markovian sampling. As a consequence, the model explains the constant-stepsize error floor through the interaction between Markovian long-run covariance and the contraction geometry of the projected Bellman operator.

Progress in AI has largely been driven by methods that assume less. As compute and data increase, approaches with weaker inductive biases generally outperform those with stronger assumptions. This is particularly characteristic of the field of Visual Representation Learning, where approaches have gone from being dominated by Supervised Learning, to Weakly Supervised Learning, to the now widespread success of Self-Supervised Learning without human labels. Yet, even modern Self-Supervised Learning approaches still depend on strong inductive biases such as augmentations, masking, or cropping. If this trend holds, even these remaining biases should become bottlenecks at scale -- and our experiments confirm this: the optimal strength of inductive biases decreases as data grows. This motivates the search for approaches that rely on fewer assumptions. To this end, we introduce Temporal Difference in Vision (TDV), a new paradigm for self-supervised learning from video that avoids existing inductive biases, relying instead on a causal assumption that the past causes the future. TDV functions by jointly training an image encoder and a motion encoder so that the current frame's representation plus the encoded motion equals the next frame's representation. Despite not leveraging any strong inductive biases, TDV matches state-of-the-art recipes on dense spatial tasks, laying the foundation for representation learning without strong assumptions.

Diffusion models are typically trained with objectives that focus on local denoising targets at individual time steps (or adjacent pairs), which do not enforce consistency between predictions along the denoising trajectory. This lack of cross-time consistency can degrade performance, especially for few-step samplers. We introduce a temporal difference (TD) objective that penalizes inconsistency of the model's multi-step progress along the denoising path. By reformulating the diffusion process as a Markov reward process and casting denoising as a policy evaluation problem in reinforcement learning, we derive a unified TD approach that applies to both discrete- and continuous-time diffusion formulations. We further propose a principled sample-based reweighting method that stabilizes training. Empirically, we show that using our TD training can significantly improve sample quality measured by FID, with stronger advantages when the number of sampling steps is small, highlighting its practical utility under low-computation-budget scenarios. We provide ablation studies to justify our design choices, including pairwise loss reweighting, regularization weight, and one-step stride. Overall, our TD approach can be a general drop-in that enforces cross-time consistency and improves generation quality across different diffusion generative models.

In this paper, we study the finite-time behavior of the TD(0) temporal-difference method with linear function approximation (LFA). We consider on-policy independent and identically distributed (i.i.d.) samples, a constant learning step, and the Polyak-Juditsky averaging method. We establish a new convergence rate, for the Mean-Square Error (MSE) on the approximated function, that is (i) fast in the sense that it admits an optimal dependency in the number of iterations k (i.e., of order 1/k), (ii) robust to ill-conditioning: it only depends on an initial error and modelindependent constants and (iii) sharp up to a multiplicative constant lower than 11. In particular, it does not depend on the smallest eigenvalue of the uncentered covariance matrix of the linear parametrization, unlike all pre-existing O(1/k) rates in the TD(0) literature. We also introduce PCTD(0), a variant of TD(0), which benefits from better convergence properties under an additional assumption of strong mixing on the Markov Chain.

This work studies the convergence of two-timescale stochastic approximations (SA), a class of iterative algorithms that update two sets of parameters in fast and slow timescales respectively. Notable examples of two-timescale SA in reinforcement learning (RL) include temporal difference learning with gradient correction (TDC) and actor-critic methods. Previously, the stability (i.e., boundedness) and convergence of two-timescale SA were only established under i.i.d. noise. This work instead establishes the stability and convergence of two-timescale SA under Markovian noise, a setup that is more realistic in RL. Notably, we do not need to use any projection operator and the noise does not need to live in a compact space. Our key technical novelty is to control the fast timescale parameter with the running max of the slow timescale parameter, instead of with the current slow timescale parameter, as most prior works do. As a key application, we establish the first almost sure convergence of TDC with eligibility traces under off-policy learning with linear function approximation.

Multimodal Large Language Models (MLLMs) face significant computational overhead when processing long videos due to the massive number of visual tokens required. To improve efficiency, existing methods primarily reduce redundancy by pruning or merging tokens based on importance or similarity. However, these approaches largely overlook a critical dimension of video content, i.e., changes and turning points, and they lack a collaborative model for spatio-temporal relationships. To address this, we propose a new perspective: similarity is for identifying redundancy, while difference is for capturing key events. Based on this, we designed a training-free framework named ST-SimDiff. We first construct a spatio-temporal graph from the visual tokens to uniformly model their complex associations. Subsequently, we employ a parallel dual-selection strategy: 1) similarity-based selection uses community detection to retain representative tokens, compressing static information; 2) temporal difference-based selection precisely locates content-changing points to preserve tokens that capture key dynamic shifts. This allows it to preserve both static and dynamic content with a minimal number of tokens. Extensive experiments show our method significantly outperforms state-of-the-art approaches while substantially reducing computational costs. Our code is available in https://github.com/bingjunluo/ST-SimDiff.

Temporal-difference learning with function approximation can be unstable under off-policy sampling. TDC stabilizes off-policy TD through an auxiliary covariance correction, and TDRC further regularizes this correction in a single-timescale recursion. This paper studies a behavior-aware replacement of the auxiliary covariance geometry in the linear prediction setting, which is the standard local model for understanding the feature-space dynamics of value-function approximation. We first replace the TDC auxiliary matrix (C) by the behavior Bellman matrix (A_μ), yielding BA-TDC, and then regularize the same behavior-aware equation to obtain BA-TDRC. This two-step construction separates the contribution of behavior-aware geometry from the contribution of regularization. The linear analysis also provides a tractable model for an auxiliary-geometry design question that arises in neural-network value approximation, where feature covariances and temporal transition matrices jointly shape the last-layer correction dynamics. We give a finite-state mean-system formulation, prove fixed-point preservation and almost-sure convergence under a Hurwitz stability condition on the instantiated mean system, and compare deterministic mean rates through the spectral radius of the exact linear error recursion. Experiments on the two-state counterexample, Baird's counterexample, Random Walk, and Boyan Chain show that the behavior-aware replacement can be highly beneficial by itself on some tasks, but that regularization is necessary for robust performance across harder settings.

Gradient temporal-difference methods provide stable off-policy prediction with linear function approximation, but their practical performance is strongly affected by the geometry induced by the auxiliary-variable metric. Existing Mirror-Prox TD methods typically use the feature covariance metric, whereas hybrid TD methods suggest that behavior-policy transition information can provide a more informative update geometry. This paper proposes a behavior-induced Mirror-Prox temporal-difference method, called STHTD-MP, which replaces the covariance metric in the primal-dual saddle-point formulation with the symmetric part of the behavior-policy Bellman matrix. The method keeps a single learning rate for the primal and auxiliary variables and applies a Mirror-Prox prediction-correction step to the resulting hybrid saddle-point operator. We provide a formal convergence analysis for fixed-policy linear prediction under standard stochastic approximation assumptions: the behavior-induced metric is positive definite, the joint mean system is Hurwitz, boundedness follows from a Lyapunov argument, and the stochastic recursion converges by the ODE method. We further derive projected-oracle ergodic gap bounds and an exact mean-operator comparison with GTD2-MP based on the spectral radius of the deterministic Mirror-Prox error matrix. The analysis shows that STHTD-MP can have a smaller mean contraction factor than GTD2-MP when the behavior-induced metric improves the saddle-point geometry. Exact numerical mean-operator analysis on two-state, Random Walk, and Boyan Chain benchmarks supports this condition, while Baird's counterexample is identified as a singular boundary case where the strict assumptions fail.

We study how temporal correlations in the data can make certain sparse learning problems efficiently learnable by gradient-based methods. Our focus is on Boolean k-juntas, a canonical sparse learning problem known to pose barriers for gradient-based methods under independent uniform samples. We show that this picture changes when the samples are generated by a lazy random walk on the hypercube. In this setting, the temporal dependencies can be exploited by a two-layer ReLU network trained using stylized-SGD with a temporal-difference loss, which compares target and predicted increments across consecutive samples. For every fixed k, the resulting sample complexity is essentially linear in the ambient dimension d. By contrast, we show that for large-batch gradient methods using standard convex pointwise losses, temporal correlations do not provide the same advantage.

A central challenge in reinforcement learning (RL) is to learn models that generalize beyond the tasks on which they are trained, a goal traditionally pursued through multi-task and meta RL. Recently, transformer architectures have emerged as a promising approach, enabling adaptation to new tasks via in-context learning without explicit parameter updates. From a functional perspective, a transformer can be viewed as a functional operator that maps a context to a task-specific function. It is thus fundamental to understand and design this operator to support stronger generalization in RL. In this work, we address this resulting question of generalization from a kernel-based perspective by establishing a connection between non-linear transformers and kernel-based temporal difference learning. By interpreting the transformer as performing regression in a Reproducing Kernel Hilbert Space (RKHS), we show that value functions from different domains can be represented using a shared set of weights, provided they lie within the same RKHS. Experiments on multiple MetaWorld domains support this interpretation, demonstrating convergence of the temporal-difference objective.

Recent work on provably efficient algorithms for learning with distribution shift has focused on two models: PQ learning (Goldwasser et al. (2020)) and TDS learning (Klivans et al. (2024)). Algorithms for TDS learning are allowed to reject a test set entirely if distribution shift is detected. In contrast, PQ learners may only reject points that are deemed out-of-distribution on an individual basis. Our main result is a surprising equivalence between these two models in the distribution-free setting. In particular, we give an efficient black-box reduction from PQ learning to TDS learning for any Boolean concept class. This equivalence implies the first hardness results for distribution-free TDS learning of basic classes such as halfspaces. The main technical contribution underlying our equivalence is a method for boosting, via branching programs, the weak distinguishing power of TDS learners that have rejected the target domain. We also show that giving a learner access to membership queries sidesteps these hardness results and allows for efficient, distribution-free PQ learnability of halfspaces. Our algorithm iteratively recovers large-margin separators obtained by applying successive Forster transforms on the training data.

Learning rate is a critical component of reinforcement learning (RL). This work uses global and local clocks to distinguish two types of learning rates. The former is of the standard form $α_t$ that depends only on the time step $t$ (i.e., a global clock). The latter is of the form $α_{ν(S_t, t)}$, where $ν(s, t)$ counts the number of visits to state $s$ until time $t$ (i.e., a local clock). In discounted RL, an RL algorithm that is convergent with a local clock is always also convergent with a global clock, and vice versa. We are not aware of any counterexample. The key contribution of this work is to show that this nice correspondence breaks down in average-reward RL. Specifically, we construct a counterexample showing that although differential temporal difference learning is convergent with a local clock, it can diverge with a global clock. This counterexample closes the open problem in Wan et al. [2021], Blaser et al. [2026].

Recent non-asymptotic analyses have substantially advanced the theory of distributional policy evaluation, but they largely concern synchronous full-state updates under a generative model, model-based estimators, accelerated variants, or different approximation architectures. Standard categorical temporal-difference learning is typically used in a different regime. It asynchronously performs a single-state update at each iteration and, in online settings, is driven by a Markovian trajectory. This leaves an important gap between existing finite-iteration theory and the categorical recursions most closely aligned with practical distributional temporal-difference implementations. We bridge this gap for two categorical policy-evaluation methods: scalar categorical temporal-difference learning in the Cramér geometry and multivariate signed-categorical temporal-difference learning in the maximum mean discrepancy geometry. After suitable isometric embeddings, both algorithms take the form of asynchronous single-state stochastic-approximation recursions that contract in a statewise supremum norm. This permits finite-iteration guarantees in discounted problems under both i.i.d. and Markovian state sampling, and in undiscounted fixed-horizon problems under i.i.d. episodic sampling.

Differential temporal difference (TD) methods are value-based reinforcement learning algorithms that have been proposed for infinite-horizon problems. They rely on reward centering, where each reward is centered by the average reward. This keeps the return bounded and removes a value function's state-independent offset. However, reward centering can alter the optimal policy in episodic problems, limiting its applicability. Motivated by recent works that emphasize the role of normalization in streaming deep reinforcement learning, we study reward centering in episodic problems and propose a generalization of differential TD. We prove that this generalization maintains the ordering of policies in the presence of termination, and thus extends differential TD to episodic problems. We show equivalence with a form of linear TD, thereby inheriting theoretical guarantees that have been shown for those algorithms. We then extend several streaming reinforcement learning algorithms to their differential counterparts. Across a range of base algorithms and environments, we empirically validate that reward centering can improve sample efficiency in episodic problems.

We formally establish the equivalence between Observation Delay (OD) and Action Delay (AD) in cooperative partially observable multi-agent systems using observation-action histories. We show that both systems generate identical admissible joint-policy sets, and their induced state-action-observation trajectories are identical in distribution, leading to identical optimal solutions in Decentralized Partially Observable Markov Decision Processes (Dec-POMDPs). This formally generalizes existing infinite-horizon single-agent results to any-horizon partially observable cooperative multi-agent problems with decentralized policy execution, and allows any mixed-delay configuration to be reduced to a pure OD system. We further prove that in Transition-Independent MDPs (TI-MDPs), the observation-action history reduces to a tractable minimal local augmented state. However, we show through numerical experiments that although the optimal solution spaces are structurally isomorphic, the practical learning dynamics are fundamentally different. First, using the minimal local augmented state, the equivalence no longer holds when transitions are not independent. Second, operational constraints and causal credit-assignment errors in Temporal Difference (TD) algorithms induce different learning behaviors across regimes. Finally, leveraging this structural equivalence to bypass these learning challenges, we demonstrate successful multi-agent zero-shot policy transfer from OD to AD, paving the way for unified, efficient solution methods in complex delayed systems.

The analysis of Temporal Difference (TD) learning in the average-reward setting faces notable theoretical difficulties because the Bellman operator is not contractive with respect to any norm. This complicates standard analyses of stochastic updates that are effective in discounted settings. Although a considerable body of literature addresses these challenges, existing theoretical approaches come with limitations. We introduce a novel algorithm designed explicitly for policy evaluation in the average-reward setting, utilizing sampling from two Markovian trajectories. Our proposed method overcomes previous limitations by guaranteeing convergence to the unique solution of a properly defined projected Bellman equation. Notably, and in contrast to earlier work, our convergence analysis is uniformly applicable to both linear function approximation and tabular settings and does not involve explicit dimension-dependent terms in its convergence bounds. These results align with what is known to hold in the discounted setting. Furthermore, our algorithm achieves improved dependence on the problem's condition number, reducing the sample complexity from quartic, as in prior literature, to quadratic scaling, and thus matching the efficiency seen in the discounted setting.

Off-policy temporal-difference (TD) learning with function approximation faces a structural tradeoff among stability, projection geometry, and variance control. Emphatic TD (ETD) improves the off-policy projection geometry through follow-on emphasis, but the follow-on trace can have high variance. We revisit this tradeoff through Bellman-error centering. Although centering naturally removes a common drift term from TD errors, we show that a naive centered emphatic extension introduces an auxiliary coupling that can destroy the positive-definiteness of the ETD key matrix. We propose \emph{Regularized Emphatic Temporal-Difference Learning} (RETD), which preserves the follow-on trace and regularizes only the auxiliary centering recursion, corresponding to lifting the lower-right block of the coupled key matrix from $1$ to $1+c$. We derive the RETD core matrix, prove convergence under a conservative sufficient regularization condition, and evaluate the method on diagnostic linear off-policy prediction tasks. The experiments show that RETD avoids the instability of naive centered emphatic learning, preserves favorable emphatic geometry, and exhibits a robust intermediate regime for the regularization parameter $c$ across the diagnostics.

Recent advances in vision-language-action (VLA) models for robotics have highlighted the importance of reliable uncertainty quantification in sequential tasks. However, assessing and improving calibration in such settings remains mostly unexplored, especially when only partial trajectories are observed. In this work, we formulate sequential calibration for episodic tasks, where task-success confidence is produced along an episode, while success is determined at the end of it. We introduce a sequential extension of the Brier score and show that, for binary outcomes, its risk minimizer coincides with the VLA policy's value function. This connection bridges uncertainty calibration and reinforcement learning, enabling the use of temporal-difference (TD) value estimation as a principled calibration mechanism over time. We empirically show that TD calibration improves performance relative to the state-of-the-art on simulated and real-robot data. Interestingly, we show that when calibrated using TD, the VLA's single-step action probabilities can yield competitive uncertainty estimates, in contrast to recent findings that employed different calibration techniques.

Temporal Predictive Coding provides a layer-local, parallelisable mechanism for learning in recurrent systems, making it an attractive candidate for online local learning on neuromorphic and edge hardware. However, its recurrent parameter update captures only local temporal relationships, neglecting the historic influence of parameters along the latent-state trajectory, and therefore struggles to assign credit over longer temporal horizons. This work combines for the first time Temporal Predictive Coding with Real-Time Recurrent Learning (tPC-RTRL), incorporating an online influence matrix that tracks this historic effect whilst preserving the spatial and temporal locality properties valued by neuromorphic implementations. Under explicit assumptions, we prove that tPC-RTRL recovers the gradients of backpropagation-through-time exactly. Empirically, a near-equivalence holds across several tasks of varying scale and complexity, including byte-level language modelling on WikiText-103 (tPC-RTRL vs. BPTT: 1.865 vs. 1.864 validation BPC), English--French translation on a CCMatrix subset (20.23 vs. 20.29 BLEU), and a realistic nanodrone system-identification benchmark (0.506m vs. 0.505m mean position error). Finally, we show that the iterative inference mechanism used during training can be reused at deployment time to incorporate intermittent state observations, halving final-position error relative to open-loop rollout on the nanodrone task (0.402m vs. 0.805m) and suggesting a path towards unifying learning and filtering within the same computational framework.

We develop a statistical theory of temporal learnability in recurrent neural networks, quantifying the maximal temporal horizon $\mathcal{H}_N$ over which gradient-based learning can recover lag-dependent structure at finite sample size $N$. The theory is built on the effective learning rate envelope $f(\ell)$, a function that captures how gating mechanisms and adaptive optimizers jointly shape the coupling between state-space dynamics and parameter updates during Backpropagation Through Time. Under heavy-tailed ($α$-stable) fluctuations, where empirical averages concentrate at rate $N^{-1/κ_α}$ with $κ_α= α/(α-1)$, the interplay between envelope decay and statistical concentration yields explicit scaling laws for the growth of $\mathcal{H}_N$: logarithmic, polynomial, and exponential temporal learning regimes emerge according to the decay law of $f(\ell)$. These results identify envelope decay as the key determinant of temporal learnability. Slower attenuation of $f(\ell)$ enlarges $\mathcal{H}_N$, while heavy-tailed fluctuations compress it by weakening statistical concentration. Moreover, envelope geometry outweighs dataset size: slowing the envelope's decay enlarges $\mathcal{H}_N$ more than adding data, so more complex architectures that realize slower-decaying envelopes can be more data-efficient than simpler ones. Experiments across multiple gated architectures and optimizers corroborate these structural predictions.

We investigate the statistical properties of Temporal Difference (TD) learning with Polyak-Ruppert averaging, arguably one of the most widely used algorithms in reinforcement learning, for the task of estimating the parameters of the optimal linear approximation to the value function. Assuming independent samples, we make three theoretical contributions that improve upon the current state-of-the-art results: (i) we establish refined high-dimensional Berry-Esseen bounds over the class of convex sets, achieving faster rates than the best known results, and (ii) we propose and analyze a novel, computationally efficient online plug-in estimator of the asymptotic covariance matrix; (iii) we derive sharper high probability convergence guarantees that depend explicitly on the asymptotic variance and hold under weaker conditions than those adopted in the literature. These results enable the construction of confidence regions and simultaneous confidence intervals for the linear parameters of the value function approximation, with guaranteed finite-sample coverage. We demonstrate the applicability of our theoretical findings through numerical experiments.