Online

In the forward reinforcement-learning problem, the reward is fixed and known; the learner is asked to find a good policy or value function. Here we turn the question around. Given offline data generated by an expert, can we recover the reward the expert was optimizing? This is the inverse reinforcement learning problem, and remarkably, two communities, structural econometricians studying dynamic discrete choice (DDC) and machine learners studying entropy-regularized IRL, have been working on exactly the same probabilistic model under different names. We begin by proving their equivalence. We then develop the classical identification result of Magnac and Thesmar and the classical computational paradigms that grew out of it: Rust's nested fixed-point algorithm, the conditional-choice-probability approach of Hotz and Miller, and the two temporal-difference approaches of Adusumilli and Eckardt: linear semi-gradient TD and approximate value iteration. Each route has its limits: dimensionality, transition-kernel estimation, the deadly triad, or projected fixed-point bias. We then walk through the modern ML/IRL strand: adversarial IRL, occupancy matching, IQ-Learn, and offline ML-IRL, deriving each method's actual objective and stating precisely what it does and does not identify. We close with the empirical-risk-minimization framework of Kang et al., which yields a gradient-based estimator for offline IRL/DDC.

Self-supervised online traversability estimation enables robots to continuously learn from unlabeled open-world experiences and adapt their navigation behavior toward safe and efficient trajectories. Existing approaches either rely on handcrafted proprioceptive traversability scores, limiting robot-agnosticism, or cluster prior data, preventing online learning. Moreover, many continual learning methods incur substantial memory and computational costs, hindering onboard deployment. We introduce COTRATE, an online learning framework for continuous traversability estimation from multimodal, unlabeled robot experience. Our method first infers robust traversability scores using a robot-agnostic, learning-based online terrain assessment module operating on proprioceptiveand inertial signals. These scores then supervise a visual traversability network through a novel alignment loss that associates visual embeddings with online terrain assessments. To mitigate forgetting during continual learning with minimal overhead, we propose a diversity-aware feature selection strategythat preserves performance using a compact replay memory. We further show that the learned traversability representation supports knowledge transfer across different robot platforms with different locomotion kinematics. We evaluate COTRATE on a dataset of $\approx$ 50,000 images collected with two robotic platforms across 11 outdoor terrains, and benchmark it on navigation tasks in three representative outdoor environments. We make the dataset, code, and trained models publicly available.

Industrial video-on-demand (VOD) recommenders need richer content understanding, but LLM-as-reranker designs repeat prompt construction, token generation, model invocation, output parsing, and fallback handling for each request. In high-volume latency-sensitive services, these request-time operations complicate throughput planning, tail-latency control, capacity isolation, and predictable operation. This paper presents Ocean4Rec, a reranking layer that uses an LLM only offline to materialize item OCEAN profiles from content metadata. Items are mapped into Openness, Conscientiousness, Extraversion, Agreeableness, and Neuroticism scores, while user profiles are built by time-decayed aggregation of recently clicked and deep-linked items in the same five-dimensional space. At request time, Ocean4Rec joins precomputed item profiles, user profiles, base recommender scores, and catalog recency, then performs numeric reranking without an LLM call. On anonymized Samsung Smart TV VOD logs, same-candidate Top1000 temporal-holdout offline evaluations show that Ocean4Rec improves NDCG@20 over a stronger non-OCEAN Base+Recency ordering by 7.6% for an NCF generator and 61.5% for a LightGCN generator. HR@20 is inconclusive for NCF and improves by 67.3% for LightGCN, reflecting sparse exact-item replay labels and the strength of recency as an industrial baseline. The result should be read as offline replay evidence for a bounded auxiliary content-taste feature that preserves the deployability advantage of a request-time-LLM-free serving path.

Ride-hailing platforms like DiDi Chuxing operate in highly dynamic environments where balancing driver supply and passenger demand is critical. Although driver-side subsidies serve as a primary lever to align these forces and improve key KPIs like completed rides (\texttt{Rides}) and gross merchandise value (\texttt{GMV}), optimizing them in production requires simultaneously meeting three constraints: (i) responsiveness to stochastic shocks, (ii) strict subsidy-rate caps, and (iii) low-latency execution at city scale. These requirements rule out expensive per-order optimization, calling for a forward-looking, constraint-aware city-level controller for online sequential decision making. To meet these requirements, we introduce D$^3$-Subsidy (Dynamic Driver-side Diffusion-based Subsidy), a hierarchical diffusion-based framework for deployable city-wide subsidy control. To bridge the train-inference gap, D$^3$-Subsidy employs a prefix-conditioned diffusion model that samples plausible future trajectories from immutable historical observations, ensuring the training protocol aligns with the fixed-history nature of online deployment. These generated plans are then decoded by a context-conditioned inverse module into low-dimensional city-level control signals. For scalable execution, we bridge the gap between city-level planning and fine-grained dispatch via a Lagrangian-dual-derived mapping, which embeds subsidy-rate caps directly into order-driver incentives without iterative optimization. Additionally, a multi-city pretraining strategy with parameter-efficient fine-tuning enables robust transfer across heterogeneous cities. Extensive offline evaluations demonstrate that D$^3$-Subsidy improves \texttt{Rides} and \texttt{GMV} while enhancing cap compliance, and a real-world A/B test confirms significant uplift while keeping budget-related violation metrics within operational thresholds.

Multi-agent LLM workflows -- systems composed of multiple role-specific LLM calls -- often outperform single-prompt baselines, but they remain difficult to debug and refine. Failures can originate from subtle errors in intermediate outputs that propagate to downstream nodes, requiring developers to inspect long traces and infer which agent to modify. We present PROTEA, a unified interface for offline, test-driven improvement of multi-agent workflows. PROTEA executes a workflow, scores intermediate node outputs with configurable rubrics, and overlays per-node states and rationales on the workflow graph to localize likely bottlenecks. To support complex systems where final-answer references are the primary supervision, PROTEA performs backward node evaluation: it generates candidate node-level expectations from final-answer references and graph context, then compares them with observed node outputs. For selected nodes, PROTEA presents targeted prompt revisions as editable before/after comparisons, then automatically reruns and re-evaluates the workflow to show output changes and score trajectories within the same interface. In two production-adjacent workflows, PROTEA improved document-inspection accuracy from 64.3% to 83.9% and recommendation Hit@5 from 0.30 to 0.38. In a formative study with six experienced LLM developers, participants valued graph-level localization, per-node rationales, and editable before/after prompt revisions.

INT2 KV-cache quantization is attractive for long-context LLM serving, but it remains difficult to make both accurate and deployable. Simple rotations such as Hadamard transforms reduce outliers, but still degrade at INT2 because they are not aligned with downstream attention. We propose OSCAR, an Ultra-low-bit KV Cache quantization method that estimates attention-aware covariance structures offline and uses them to derive fixed rotations and clipping thresholds for quantization. In this way, it aligns KV quantization with the covariance structures that attention actually consumes. More importantly, we not only provide theoretical justification but also develop a fully deployable OSCAR system with a custom INT2 attention kernel that remains compatible with paged KV-cache serving and fused kernel pipelines, enabling seamless integration into modern LLM serving frameworks such as SGLang and vLLM. We evaluate our methods on recent reasoning models with reasoning traces of up to 32k tokens across 5 tasks. On Qwen3-4B-Thinking-2507 and Qwen3-8B, OSCAR reduces the BF16 accuracy gap to 3.78 and 1.42 points, respectively, while naive rotation INT2 collapses to nearly zero. We further scale OSCAR to Qwen3-32B and GLM-4.7 (358B params), where it remains effectively on par with BF16. On long context - RULER-NIAH up to 128K, OSCAR remains robust on both Qwen3 models, while naive rotation INT2 collapses. System-wise, OSCAR reduces KV-cache memory by approximately 8x, improves throughput by up to 7x at large batch sizes under the same memory budget, and accelerates batch-size-1 decoding by up to 3x over BF16 due to reduced memory bandwidth overhead.

Security systems demand continuous, cryptograph- ically robust identity verification without requiring subjects to carry physical tokens, smart cards, or dedicated hardware authenticators. This paper presents BIDO (Biometric Identity Online), a device-free authentication standard that achieves Au- thenticator Assurance Level 2 (AAL2) per NIST SP 800-63B with- out storing long-lived biometric templates, facial images, or any other form of Personally Identifiable Information (PII). BIDO derives Elliptic Curve Digital Signature Algorithm (ECDSA) key material deterministically from a live biometric measurement salted with a user-defined memorized secret at every authen- tication event, eliminating persistent private-key storage while enabling verification from any commodity sensor terminal. The generated credentials are non-discoverable (non-resident) Web Authentication (WebAuthn) credentials, fully compatible with all FIDO2-enabled websites and services without modification on the server side. A multi-stage pipeline, comprising capture of 200 valid biometric samples, feature extraction using the Dlib 68- point facial landmark predictor, affine face alignment, frontality gating, Euclidean distance computation from the inter-eye mid- point, floor-division quantization with divisor q = 8, inter-session drift stabilization, and majority-voting SHA-256 hash binding, produces a Verification Seed (Vseed) from which the WebAuthn credential is transiently derived and immediately zeroized after signing. Evaluated against three prominent face benchmarks (VGGFace2, LFW, and MegaFace), achieving 99.51% verification accuracy on LFW and 92.14% Rank-1 identification accuracy on MegaFace Challenge 1 at 10^6 distractors, with a cryptographic False Accept Rate (FAR) of 0.03%, a False Reject Rate (FRR) of 0.90%.

During large-scale evacuations, concentrated electric vehicle (EV) charging demand can overload fixed charging stations (FCSs), leading to prolonged waiting time and increased risk exposure. To address this challenge, this study proposes dynamically deploying mobile charging trucks (MCTs) to complement FCSs, and develops an Adaptive Risk-aware MCT Deployment (ARMD) framework for real-time operation. It divides the MCT deployment into two problems: risk-aware allocation of MCTs among FCSs and dynamic routing of MCTs to the assigned FCSs, and solves them under an offline-to-online paradigm. The resource allocation problem is formulated as a decentralized partially observable Markov decision process, and a multi-agent proximal policy optimization (MAPPO)-based policy is developed to coordinate multiple MCTs under decentralized observations. The policy is pre-trained offline in an evacuation simulator and adaptively refined online according to current evacuation context. For routing, a spatio-temporal travel time predictor is developed to support rolling-horizon route updates. The proposed framework is evaluated in a simulated hurricane evacuation environment built using real-world data from Hillsborough County, Florida. Experiments show that ARMD consistently outperforms offline optimization, online heuristic dispatch, and rolling-horizon optimization in reducing risk exposure. For demand perturbation scenarios, ARMD reduces average risk exposure by up to 71.1%, relative to the baseline without MCTs. In the case of fixed e-vehicle charging infrastructure or road link failures, ARMD achieves 39.3% to 60.5% reduction in average risk exposure, with its advantages becoming more pronounced as the severity of disruption increases. These results demonstrate the effectiveness and robustness of ARMD in enhancing mobile charging operations for realistic scenarios of uncertain evacuation conditions.

Online OS tuning can improve long-running services, but existing controllers are poorly matched to live hosts. They treat scheduler, power, memory, and I/O controls as black-box variables and optimize a scalar reward. This view ignores cross-knob policy structure, breaks down when application metrics are unavailable, and can send a running service into degraded regions that persist after the bad setting is removed. We present TuxBot, a host-side framework for steady-state OS tuning with bounded language-model guidance. TuxBot turns knob schemas, telemetry, current configuration, recent action--response history, and retrieved prior runs into a compact decision context. A fast loop proposes low-latency updates, a slower loop periodically revises the search strategy, and every proposed change passes through typed validation before reaching kernel or sysctl interfaces. This lets the controller reason about OS-control meaning and indirect performance signals while keeping model cost, latency, and authority constrained. We evaluate TuxBot on 13 live workloads from five benchmark suites while tuning up to 41 Linux parameters. Across the suite, TuxBot improves stable-phase performance by 72.5% over default settings and by 153.3% relative to the strongest non-LLM baseline. A 30-window session costs about $0.20 in model calls. With only host-level metrics, TuxBot still outperforms baselines given direct application objectives by 93.7 percentage points, while avoiding severe degraded regions reached by structure-blind exploration.

Cross-domain offline reinforcement learning aims to adapt a policy from a source domain to a target domain using only pre-collected datasets, where environment dynamics may differ. A key challenge is to leverage source data while reducing distributional mismatch, particularly when the target dataset is extremely limited. To address this, we propose Target-aligned Coverage Expansion (TCE), a framework that decides how source data should be used, either by directly incorporating target-near transitions or by expanding state coverage through target-aligned generation, guided by theoretical analysis. TCE builds on a dual score-based generative model to synthesize target-consistent transitions over an expanded state region. Extensive experiments across diverse cross-domain environments show that TCE consistently outperforms state-of-the-art cross-domain offline RL baselines.

With recent advances in embodied agents and AR devices, egocentric observations are readily available as input for real-world interactive online applications. However, egocentric viewpoints can only sporadically observe hands, in addition to the estimated head trajectory. We propose EgoForce, an online framework for reconstructing long-term full-body motion from noisy egocentric input. While existing generative frameworks can robustly handle noisy and sparse measurements, they assume a fixed-length observation window is available and are thus not suitable for real-time applications. Faster inference often relies on autoregressive prediction, sacrificing robustness. In contrast, we adopt a diffusion-based method with a temporally asymmetric noise schedule inspired by Diffusion Forcing. Specifically, our approach models temporally evolving uncertainty and incrementally denoises states as new streaming observations arrive. Combined with a noise-robust imputation strategy, EgoForce progressively generates stable and coherent full-body motion under strict causal constraints. Experiments demonstrate that our online framework outperforms existing online and offline methods, enabling long-horizon, full-body motion reconstruction in challenging egocentric scenarios.

Geometric estimation including depth estimation and scene reconstruction is a crucial technique for colonoscopy which can provide surgeons with 3D spatial perception and navigation. However, geometric ground truth in colonoscopy is difficult to obtain due to narrow and enclosed space of the colon, while there is a large feature gap between simulated data and realistic data caused by artifacts and illumination. In this paper, we present CoGE, a novel framework for online monocular geometric estimation during colonoscopy. Firstly, we propose an illumination-aware supervision module based on the Retinex theory to address illumination diversity in different colonoscopy scenes. Moreover, a structure-aware perception module is proposed based on wavelet decomposition to extract common structural and local features of the colon. Both quantitative and qualitative results demonstrate that the proposed model solely trained on simulated data achieves state-of-the-art performance in geometric estimation for both simulated and realistic scenes.

We study the problem of learning Nash equilibria in offline two-player zero-sum Markov games. While existing approaches often rely on explicit pessimism to address distribution shift, we show that KL regularization alone suffices to stabilize learning and guarantee convergence. We first introduce Regularized Offline Sequential Equilibrium (ROSE), a theoretical framework that achieves a fast $\widetilde{\mathcal{O}}(1/n)$ convergence rate under \textit{unilateral concentrability}, improving over the standard $\widetilde{\mathcal{O}}(1/\sqrt{n})$ rates in unregularized settings. We then propose Sequential Offline Self-play Mirror Descent (SOS-MD), a practical model-free algorithm based on least-squares value estimation and iterative self-play updates. We prove that the last iterate of SOS-MD attains the same $\widetilde{\mathcal{O}}(1/n)$ statistical rate up to a vanishing optimization error of order $\widetilde{\mathcal{O}}(1/\sqrt{T})$ in the number of self-play iterations $T$.

Bayesian optimization (BO) is a widely used framework for optimizing expensive black-box functions, commonly based on Gaussian process (GP) surrogate models. Its effectiveness relies on uncertainty quantification that is both sharp (informative) and well-calibrated along the BO trajectory. In practice, GP kernel hyperparameters are unknown and are refit online from sequentially collected (non-i.i.d.) data, which can yield miscalibrated or overly conservative uncertainty and lies outside the fixed-kernel assumptions of standard BO regret theory. We propose Online Sharp-Calibrated Bayesian Optimization (OSCBO), a BO algorithm that adaptively balances GP sharpness and calibration by casting hyperparameter selection as a constrained online-learning problem. We also show that OSCBO preserves sublinear regret bounds by leveraging the theoretical guarantees of the underlying online learning algorithm. Empirically, OSCBO performs competitively across synthetic and real-world benchmarks, ranking among the strongest methods in final simple regret while maintaining robust cumulative-regret behavior.

Coding harnesses such as Claude Code and OpenHands wrap foundation models with tools, memory, and planning, but no equivalent exists for embodied agents' long-horizon partial-observability decision-making. We first report our Gemini Plays Pokemon (GPP) experiments. With iterative human-in-the-loop harness refinement, GPP became the first AI system to complete Pokemon Blue, Yellow Legacy on hard mode, and Crystal without a lost battle. In the hardest stages, the agent itself began iterating on its strategy through long-context memory, surfacing emergent self-improvement signals alongside human-in-the-loop refinement. Continual Harness removes the human fully from this loop: a reset-free self-improving harness for embodied agents that formalizes and automates what we observed. Starting from only a minimal environment interface, the agent alternates between acting and refining its own prompt, sub-agents, skills, and memory, drawing on any past trajectory data. Prompt-optimization methods require episode resets; Continual Harness adapts online within a single run. On Pokemon Red and Emerald across frontier models, Continual Harness starting from scratch substantially reduces button-press cost relative to the minimalist baseline and recovers a majority of the gap to a hand-engineered expert harness, with capability-dependent gains, despite starting from the same raw interface with no curated knowledge, no hand-crafted tools, and no domain scaffolding. We then close the loop with the model itself: an online process-reward co-learning loop, in which an open-source agent's rollouts through the refining harness are relabeled by a frontier teacher and used to update the model, drives sustained in-game milestone progress on Pokemon Red without resetting the environment between training iterations.

Online Temporal Action Localization (On-TAL) aims to detect the occurrence time and category of actions in untrimmed streaming videos immediately upon their completion. Recent advancements in this field focus on developing more sophisticated frameworks, shifting from Online Action Detection (OAD)-based aggregation paradigm to instance-level understanding. However, existing approaches are typically trained on specific domains and often exhibit limited generalization capabilities when applied to arbitrary videos, particularly in the presence of previously unseen actions. In this paper, we introduce a new task called Online Zero-shot Temporal Action Localization (OZ-TAL), which aims to detect previously unseen actions in an online fashion. Furthermore, we propose a training-free framework that leverages off-the-shelf Vision-Language Models (VLMs) while introducing additional mechanisms to enhance visual representations and mitigate their inherent biases. We establish new benchmarks and representative baselines for OZ-TAL on THUMOS14 and ActivityNet-1.3, and extensive experiments demonstrate that our method substantially outperforms existing state-of-the-art approaches under both offline and online zero-shot settings.

We consider the problem of learning an unknown subset $N_\text{target}$ of a domain in an online setting. In each round $t$, the learner predicts a set of items ${N}_t$ and receives one of two types of feedback, each with equal probability: precision feedback, in which a randomly chosen item from the predicted set $N_t$ is revealed and the learner is told whether it belongs to $N_\text{target}$ (incurring a reward if it does), or recall feedback, in which a randomly chosen item from the target set $N_\text{target}$ is revealed and the learner is told whether it belongs to $N_t$ (incurring a reward if it does). The goal is to maximize the cumulative reward over time. This simple online set learning problem abstracts a variety of learning scenarios with precision- and recall-type feedback. We show that a hypothesis class (a family of subsets of the domain) is learnable in this setting if and only if it has finite Vapnik-Chervonenkis (VC) dimension, mirroring the classical PAC characterization. However, the resulting algorithmic structure is markedly more intricate: in contrast to standard Probably Approximately Correct (PAC) learning -- where the algorithmic landscape is governed by the simple principle of Empirical Risk Minimization (ERM) -- our partial feedback model can invalidate ERM and even all proper learning rules. We develop algorithms to address the dependencies induced by the feedback, obtaining regret guarantees in both the realizable and agnostic settings. Our results provide a qualitative characterization of learnability in this model, addressing its most basic question, while pointing to a range of natural and intriguing open questions, including the determination of optimal regret rates.

We study online estimation in latent-variable models by recasting the problem as tracking a moving empirical equilibrium. Standard online EM and stochastic approximation analyses primarily study convergence toward the population parameter and typically do not isolate the empirical batch optimum from the online tracking error at finite horizon. Our framework decomposes the online estimate into the frozen batch equilibrium at the current running statistic and a tracking lag that captures the algorithm's delay behind this moving target. We prove a batch-to-online transfer theorem: provided $\lVert e_T \rVert_{L^{2}} = o(T^{-1/2})$, the online estimator inherits the batch central limit theorem and the sharp first-order risk constant. Our key observation is that the empirical optimum evolves on a smooth equilibrium manifold indexed by the running statistic. An $m$-th order equilibrium-jet predictor combined with an order-$ν$ frozen corrector yields localized tracking rates $O(T^{-ν(m+1)})$. We formalize EM-compressibility and EM-jet$^R$-compressibility as the structural conditions that make the equilibrium response and the Newton corrector evaluable from a retained streaming statistic. The theory is instantiated in latent linear Gaussian covariance estimation, where the first-order scheme operates on a compressed $d \times d$ statistic with explicit finite-sample risk envelopes and a certified restart rule.

Many real-world resource allocation systems, such as humanitarian logistics and vaccine distribution, must preposition limited supply across multiple locations before demand is realized while stockouts incur irreversible service losses. To study this, we introduce the Online Shared Supply Allocation (OSSA) problem, a stateful online model in which a central hub allocates a finite, unknown supply to multiple sites facing sequential demand under fixed-charge transportation costs and lost-sales penalties. Unlike classical make-to-stock or make-to-order inventory models, OSSA precludes backlogging and replenishment only hedges against future demand. To tackle OSSA, we propose a deterministic threshold-proportional policy GPA and prove that it achieves a $4/3$-approximation to the offline optimum up to an additive term independent of the total supply. We complement this with matching lower bounds showing that the $4/3$ ratio is tight and that the additive-error dependence is unavoidable, even for randomized algorithms that know the total supply upfront. Finally, we develop a learning-augmented extension to GPA that principally incorporates imperfect forecasts (e.g., from human experts or ML models) commonly available in practice, enabling us to exploit high-quality advice while being robust against arbitrary bad ones. Synthetic and real-world experiments show that GPA outperforms natural baselines with global supply is scarce.

Backdoor attacks pose a serious threat to deep reinforcement learning (DRL). Current defenses typically rely on reward anomalies to reverse-engineer triggers and model finetuning to remove backdoors. However, complex trigger patterns undermine their robustness, and fine-tuning entails high costs, limiting practical utility. Therefore, we shift defense concerns to trigger-agnostic backdoor output behaviors and propose BehaviorGuard, an online behavior-based backdoor detection and mitigation framework for DRL. Specifically, we find that regardless of attacks, backdoored policies induce consistent shifts in action distributions to ensure reliable activation, leaving detectable traces in high-quantile regions and distribution tails, even in the absence of triggers. Based on this, we design a novel metric that captures behavioral drift in action distributions to identify and suppress backdoor actions at runtime. To our knowledge, this is the first online backdoor defense that counters attacks both in single- and multi-agent DRL. Evaluated across diverse benchmarks with different backdoor attacks, BehaviorGuard consistently surpasses prior methods in both efficacy and efficiency.

Product reviews significantly influence purchasing decisions on e-commerce platforms. However, the sheer volume of reviews can overwhelm users, obscuring the information most relevant to their specific needs. Current e-commerce summarization systems typically produce generic, static summaries that fail to account for the fact that (i) different users care about different product characteristics, and (ii) these preferences may evolve with interactions. To address the challenge of unknown latent preferences, we propose an online learning framework that generates personalized summaries for each user. Our system iteratively refines its understanding of user preferences by incorporating feedback directly from the generated summaries over time. We provide a case study using the Amazon Reviews'23 dataset, showing in controlled simulations that online preference learning improves alignment with target user interests while maintaining summary quality.

Contextual MDPs are powerful tools with wide applicability in areas from biostatistics to machine learning. However, specializing them to offline datasets has been challenging due to a lack of robust, theoretically backed methods. Our work tackles this problem by introducing a new approach towards adaptive estimation and cost optimization of contextual MDPs. This estimator, to the best of our knowledge, is the first of its kind, and is endowed with strong optimality guarantees. We achieve this by overcoming the key technical challenges evolving from the endogenous properties of contextual MDPs; such as non-stationarity, or model irregularity. Our guarantees are established under complete generality by utilizing the relatively recent and powerful statistical technique of $T$-estimation (Baraud, 2011). We first provide a procedure for selecting an estimator given a sample from a contextual MDP and use it to derive oracle risk bounds under two distinct, but nevertheless meaningful, loss functions. We then consider the problem of determining the optimal control with the aid of the aforementioned density estimate and provide finite sample guarantees for the cost function.

Kullback-Leibler (KL) regularization is widely used in offline decision-making and offers several benefits, motivating recent work on the sample complexity of offline learning with respect to KL-regularized performance metrics. Nevertheless, the exact sample complexity of KL-regularized offline learning remains largely from fully characterized. In this paper, we study this question in the setting of multi-armed bandits (MABs). We provide a sharp analysis of KL-PCB (Zhao et al., 2026), showing that it achieves a sample complexity of $\tilde{O}(ηSAC^{π^*}/ε)$ under large regularization $η= \tilde{O}(ε^{-1})$, and a sample complexity of $\tildeΩ(SAC^{π^*}/ε^2)$ under small regularization $η= \tildeΩ(ε^{-1})$, where $η$ is the regularization parameter, $S$ is the number of contexts, $A$ is the number of arms, $C^{π^*}$ policy coverage coefficient at the optimal policy $π^*$, $ε$ is the desired sub-optimality, and $\tilde{O}$ and $\tildeΩ$ hide all poly-logarithmic factors. We further provide a pair of sharper sample complexity lower bounds, which matches the upper bounds over the entire range of regularization strengths. Overall, our results provide a nearly complete characterization of offline multi-armed bandits with KL regularization.

Offline goal-conditioned RL (GCRL) learns goal-reaching policies from static datasets, but real-world datasets are often partially observable and history-dependent, exhibiting a mix of Markovian and non-Markovian that violate standard RL assumptions. History-aware sequence models such as Decision Transformer (DT) are a natural fit for long-term dependency modeling, yet pure attention is inefficient and brittle when handling local Markovian structure and long-range context simultaneously. Although recent hybrid architectures (e.g., LSDT) introduce local extractors to improve local dependencies modeling, the fixed-window extraction cannot adapt its effective memory to varying dependency lengths in temporally heterogeneous settings, often truncating long-range context rather than compressing its content adaptively. Moreover, sequential offline GCRL faces a key bottleneck: under sparse rewards, return-to-go (RTG) becomes non-discriminative across sub-trajectories, providing little guidance signal for stitching goal-reaching behaviors from diverse demonstrations. To address these, we propose \textbf{QHyer}, which replaces RTG with a flow-parameterized, state-conditioned goal-reaching Q-estimator to support stitching across demonstrations, and introduces a gated Hybrid Attention-Mamba backbone that performs content-adaptive history compression while preserving local dynamics. Extensive experiments demonstrate that \textbf{QHyer} achieves state-of-the-art performance on both non-Markovian and Markovian datasets, validating its effectiveness for diverse scenarios.

Generative models have emerged as a promising paradigm for offline multi-agent reinforcement learning (MARL), but existing approaches require many iterative sampling steps. Recent few-step acceleration methods either distill a joint teacher into independent students or apply averaged velocity fields independently to each agent. Unfortunately, these few-step approaches hurt inter-agent coordination. We show that the efficiency-coordination trade-off is not inherent: single-pass multi-agent generation can preserve coordination when the velocity field is natively joint-coupled. We propose Coordinated few-step Flow (CoFlow), an architecture that combines Coordinated Velocity Attention (CVA) with Adaptive Coordination Gating. A finite-difference consistency surrogate further replaces memory-prohibitive Jacobian-vector product backpropagation through the averaged velocity field with two stop-gradient forward passes. Across 60 configurations spanning MPE, MA-MuJoCo, and SMAC, CoFlow matches or surpasses Gaussian policies, value-based methods, transformer policies, diffusion models, and prior flow baselines on episodic return. Three independent coordination probes confirm that CoFlow's improvements arise from inter-agent coordination rather than per-agent capacity. A denoising-step sweep shows that single-pass inference suffices on every configuration. CoFlow reaches state-of-the-art coordination quality in 1-3 denoising steps under both centralized and decentralized execution. Project Page: https://guowei-zou.github.io/coflow/

Offline evaluation of language models from usage logs is biased when model choice is confounded: the same user-side factors that influence which model is used can also influence how its output is judged, so raw comparisons of logged scores mix self-selected populations rather than estimating a common quantity of interest. A small randomized experiment can break this bias by overriding model choice, but in practice such experiments are scarce and costly. We study a three-source design that combines a large confounded observational log (OBS) for scale, a small randomized experiment (EXP) for unconfounded scoring, and an offline simulator (SIM) that replays candidate models on cached contexts. Our main result is an identification theorem showing that the randomized experiment and the simulator are together enough to recover causal model values; the observational log enters only afterward, to reduce estimation error rather than to make the causal comparison valid. Six estimator families are evaluated in a controlled semi-synthetic validation and in two real-task cached benchmarks for summarization and coding. No family dominates every regime; relative performance depends on the amount of unbiased EXP supervision and on how closely the target reward aligns with OBS-derived structure.

Robots in homes must continuously sense the people around them, yet most prior work relies on limited or offline perception. We argue that perception quality is the dominant factor governing what interaction is achievable at home, and build a testbed to test this claim. OmniRobotHome instruments a furnished home with 48 hardware-synchronized cameras and three manipulators in a unified world frame, delivering real-time markerless full-body human pose, 6D object pose, anticipatory motion forecasting, and a social avatar agent that converses with residents. Using the platform, we treat perception quality as an experimental variable across safety, human assistance, and social interaction, and find that interaction quality degrades measurably as real-timeness, granularity, coverage, accuracy, forecasting, or memory is weakened. All code and data will be released.

We present RADIO-ViPE (Reduce All Domains Into One -- Video Pose Engine), an online semantic SLAM system that enables geometry-aware open-vocabulary grounding, associating arbitrary natural language queries with localized 3D regions and objects in dynamic environments. Unlike existing approaches that require calibrated, posed RGB-D input, RADIO-ViPE operates directly on raw monocular RGB video streams, requiring no prior camera intrinsics, depth sensors, or pose initialization. The system tightly couples multi-modal embeddings -- spanning vision and language -- derived from agglomerative foundation models (e.g., RADIO) with geometric scene information. This coupling takes place in initialization, optimization and factor graph connections to improve the consistency of the map from multiple modalities. The optimization is wrapped within adaptive robust kernels, designed to handle both actively moving objects and agent-displaced scene elements (e.g., furniture rearranged during ego-centric session). Experiments demonstrate that RADIO-ViPE achieves state-of-the-art results on the dynamic TUM-RGBD benchmark while maintaining competitive performance against offline open-vocabulary methods that rely on calibrated data and static scene assumptions. RADIO-ViPE bridges a critical gap in real-world deployment, enabling robust open-vocabulary semantic grounding for autonomous robotics and unconstrained in-the-wild video streams. Project page: https://be2rlab.github.io/radio_vipe

Large Language Models (LLMs) have revolutionized the field of natural language processing. However, they exhibit some limitations, including a lack of reliability and transparency: they may hallucinate and fail to provide sources that support the generated output. Retrieval-Augmented Generation (RAG) was introduced to address such limitations in LLMs. One popular implementation, RAG-as-a-Service (RaaS), has shortcomings that hinder its adoption and accessibility. For instance, RaaS pricing is based on the number of submitted prompts, without considering whether the prompts are enriched by relevant chunks, i.e., text segments retrieved from a vector database, or the quality of the utilized chunks (i.e., their degree of relevance). This results in an opaque and less cost-effective payment model. We propose Chunk-as-a-Service (CaaS) as a transparent and cost-effective alternative. CaaS includes two variants: Open-Budget CaaS (OB-CaaS) and Limited-Budget CaaS (LB-CaaS), which is enabled by our ``Utility-Cost Online Selection Algorithm (UCOSA)''. UCOSA further extends the cost-effectiveness and the accessibility of the OB-CaaS variant by enriching, in an online manner, a subset of the submitted prompts based on budget constraints and utility-cost tradeoff. Our experiments demonstrate the efficacy of the proposed UCOSA compared to both offline and relevance-greedy selection baselines. In terms of the performance metric-the number of enriched prompts (NEP) multiplied by the Average Relevance (AR)-UCOSA outperforms random selection by approximately 52% and achieves around 75% of the performance of offline selection methods. Additionally, in terms of budget utilization, LB-CaaS and OB-CaaS achieve higher performance-to-budget ratios of 140% and 86%, respectively, compared to RaaS, indicating their superior efficiency.

Offline zero-shot reinforcement learning (RL) aims to learn agents that optimize unseen reward functions without additional environment interaction. The standard approach to this problem trains task-conditioned policies by sampling task vectors that define linear reward functions over learned state representations. In most existing algorithms, these task vectors are randomly sampled, implicitly assuming this adequately captures the structure of the task space. We argue that doing so leads to suboptimal zero-shot generalization. To address this limitation, we propose extracting task vectors directly from the offline dataset and using them to define the task distribution used for policy training. We introduce a simple and general reward function extraction procedure that integrates into existing offline zero-shot RL algorithms. Across multiple benchmark environments and baselines, our approach improves zero-shot performance by an average of 20%, highlighting the importance of principled task sampling in offline zero-shot RL.

Many bandit systems are deployed with offline historical data, such as past logs from earlier policies. Using these data can reduce early online exploration when they remain informative for the online problem. When the offline and online environments differ, such data can be biased for the online problem. For linear (contextual) bandits, this bias is directional: offline data may be informative in some feature directions and misleading in others. However, prior work typically controls this gap through a known Euclidean bound on the model parameters, which we prove is too coarse: even with the offline parameter known, bias in a single unknown direction can force dimension-dependent regret. To address this challenge, we introduce a directional bias certificate $(M_{\mathrm{bias}},ρ)$ that measures the offline-to-online gap through an $M_{\mathrm{bias}}$-induced norm and assigns different bias budgets to different directions. Building on this certificate, we propose \emph{Ellipsoidal-MINUCB}, which augments the online learning with an offline-pooled branch that safely exploits historical data. When the certificate is known, we show that the algorithm matches the standard SupLinUCB rate in the worst case and improves when offline coverage aligns with low-bias directions. When the certificate is unknown, we estimate it adaptively from offline and accumulated online data and establish a corresponding regret guarantee. Numerical experiments support the theory and show gains in aligned regimes.

Building scalable and reusable multi-agent decision policies from offline datasets remains a challenge in offline multi-agent reinforcement learning (MARL), as existing methods often rely on fixed observation formats and action spaces that limit generalization. In contrast, large language models (LLMs) offer a flexible modeling interface that can naturally accommodate heterogeneous observations and actions. Motivated by this, we propose the Decision Language Model (DLM), which formulates multi-agent decision making as a dialogue-style sequence prediction problem under the centralized training with decentralized execution paradigm. DLM is trained in two stages: a supervised fine-tuning phase, which leverages dialogue-style datasets for centralized training with inter-agent context and generates executable actions from offline trajectories, followed by a group relative policy optimization phase to enhance robustness to out-of-distribution actions through lightweight reward functions. Experiments on multiple benchmarks show that a unified DLM outperforms strong offline MARL baselines and LLM-based conversational decision-making methods, while demonstrating strong zero-shot generalization to unseen scenarios across tasks.

Reasoning language models can solve increasingly complex tasks, but struggle to produce the calibrated confidence estimates necessary for reliable deployment. Existing calibration methods usually depend on labels or repeated sampling at inference time, making them impractical in many settings. We introduce a method for unsupervised confidence calibration of reasoning LLMs when only a single generation is available at inference time. Our approach uses offline sampling on unlabeled data to derive a self-consistency-based proxy target, then distills this signal into a lightweight deployment-time confidence predictor. In a broad evaluation across 5 math and question-answering tasks using 9 reasoning models, our method substantially outperforms baselines, including under distribution shift, and improves downstream performance in selective prediction and simulated downstream decision-making.

Online federated learning (OFL) has emerged as a popular framework for decentralized decision-making over continuous data streams without compromising client privacy. However, the adversary model assumed in standard OFL typically precludes any potential benefits of parallelization. Further, it fails to adequately capture the different sources of statistical variation in OFL problems. In this paper, we extend the OFL paradigm by integrating a stochastically extended adversary (SEA). Under this framework, the loss function remains fixed across clients over time. However, the adversary dynamically and independently selects the data distribution for each client at each time. We propose the \algoOFL{} algorithm to solve this problem, which utilizes online stochastic gradient descent at the clients, along with periodic global aggregation via the server. We establish bounds on the global network regret over a time horizon $T$ for two classes of functions: (1) for smooth and convex losses, we prove an $\mathcal{O}(\sqrt{T})$ bound, and (2) for smooth and strongly convex losses, we prove an $\mathcal{O}(\log T)$ bound. Through careful analysis, we quantify the individual impact of both spatial (across clients) and temporal (over time) data heterogeneity on the regret bounds. Consequently, we identify a regime of mild temporal variation (relative to stochastic gradient variance), where the network regret improves with parallelization. Hence, in the SEA setting, our results improve the existing pessimistic worst-case results in online federated learning.

We propose an algorithmic framework, Offline Estimation to Decisions (OE2D), that efficiently reduces contextual bandit learning with general reward function approximation to offline regression. The framework allows near-optimal regret for contextual bandits with large action spaces with $O(\log T)$ calls to an offline regression oracle over $T$ rounds, and makes $O(\log\log T)$ calls when $T$ is known. The design of OE2D algorithm generalizes Falcon~\citep{simchi2022bypassing} and its linear reward version~\citep[][Section 4]{xu2020upper} in that it finds an action distribution that we term ``exploitative F-design'' that simultaneously guarantees low regret and good coverage, striking a balance between exploration and exploitation. Central to our regret analysis is a new complexity measure, the Decision-Offline Estimation Coefficient (DOEC), which we show is small in many settings, including bounded Eluder dimension per-context and the smoothed regret setting. We also establish a relationship between DOEC and Decision Estimation Coefficient (DEC)~\citep{foster2021statistical}, bridging the design principles of offline- and online-oracle efficient contextual bandit algorithms for the first time.

Online handwritten character recognition leverages stroke order and dynamic features, which generally provide higher accuracy and robustness compared with offline recognition. However, in practical applications, rotational deformations can disrupt the spatial layout of strokes, substantially reducing recognition accuracy. Extracting rotation-invariant features therefore remains a challenging open problem. In this work, we employ the Sliding Window Path Signature (SW-PS) to capture local structural features of characters, and introduce the lightweight Linear Recurrent Units (LRU) as the classifier. The LRU combine the fast incremental processing capability of recurrent neural networks (RNN) with the efficient parallel training of state space models (SSM), while reliably modelling dynamic stroke characteristics. We conducted recognition experiments with random rotation angle up to $\pm 180^{\circ}$ on three subsets of the CASIA-OLHWDB1.1 dataset: digits, English upper letters, and Chinese radicals. The accuracies achieved after ensemble learning were $99.62\%$, $96.67\%$, and $94.33\%$, respectively. Experimental results demonstrate that the proposed SW-PS+LRU framework consistently surpasses competing models in both convergence speed and test accuracy.

Supervised fine-tuning (SFT) datasets are critical to the downstream performance of large language models, yet they often contain low-quality or harmful question-response pairs. To improve SFT data quality, we develop a unified bilevel framework that combines offline data selection with the online self-refining generation. In the offline setting, bilevel data selection (BDS) selects question-response pairs from the offline SFT dataset to maximize the validation performance. We theoretically show that the optimal model given by BDS outperforms direct data mixing approach in useful data coverage. Moreover, we provide a global convergence analysis for gradient-based BDS approach for one-layer Transformer, showing that the epsilon-global optimum of offline BDS is achievable in finite time. Although efficient, offline BDS discards potentially harmful questions together with responses, thereby reducing question diversity. We address this limitation by refining the responses to selected questions using online self-refining generation framework. However, BDS is inefficient to update the response weights when responses are regenerated online. To address this issue, we introduce bilevel multi-objective optimization (BMO) for response-level weighting. We show that BMO recovers the same validation-aligned solution as BDS, but admits a closed-form importance-ratio weight that adapts to regenerated responses. Experiments on LLM quality enhancement and safety-aware fine-tuning demonstrate that the proposed framework consistently improves both data quality and downstream fine-tuning performance.

The performance of online mirror descent depends critically on the geometry induced by its mirror map, yet standard algorithms largely rely on two canonical choices: Euclidean and entropic geometry. We show that these two geometries can both be substantially suboptimal when loss gradients are sparse. We introduce a family of randomized block-norm mirror maps that interpolates between Euclidean and entropic geometries and adapts to intermediate sparsity structure. For several standard convex sets, including $\ell_p$ balls, ellipsoids, boxes, and Minkowski sums of norm balls, we prove polynomial-in-dimension improvements in regret bounds over the better of online projected gradient descent and exponentiated gradient. We further construct explicit online convex optimization instances for which these improvements are realized: on a simple polytope, an intermediate block geometry achieves a $\text{poly}(d)$ separation in regret from both Euclidean and entropic geometries in dimension $d$, while on the probability simplex we obtain a separation of order $\Omega(\sqrt{\log d}/\log\log d)$. Finally, we study geometry selection when sparsity is unknown. We show that naively alternating between mirror maps can incur linear regret, even though either mirror map alone has sublinear regret, and give a Hedge meta-algorithm that competes with the best mirror map in a finite portfolio. For random block geometries, this yields regret within an $O(\sqrt{\log\log d})$ factor of the best random uniform block norm chosen in hindsight.