Papers

Studying event memory requires defining events and their relationships within remembered experience. Autobiographical narratives provide evidence of this organization, but their analytical units must first be established. This study introduces a large language model framework for constructing event representations from narratives and modeling their structure and dynamics with excerpt-level traceability. Explicit boundary rules and a feature schema informed by hippocampal models of episodic memory and navigation guide event construction. In a proof-of-concept application to interviews from five participants, the framework constructed 256 event-states. Multiple correspondence analysis embedded these states in feature-based spaces, with analyses focused on a selected 4-D geometry and 10-D and 13-D reference geometries. Consecutive event-states were significantly closer than expected under randomized narrative order. In the selected 4-D ten-basin solution, similarity between basin-state feature profiles corresponded with similarity between one-step routing profiles; context, mechanism, and theme differentiated full-system pathway engagement. A successor representation modeled expected basin-state occupancy across successive steps. Correlations between basin-centroid distances and complete successor-profile distances were positive across all five basin solutions, with exploratory tests supporting correspondence in at least two solutions across all three geometries. These findings connect feature-based event representations with observed narrative transitions and modeled multistep organization. The framework translates hippocampal principles into source-linked event boundaries and state representations, providing an analytical basis for testing neural accounts of event segmentation and predictive organization during naturalistic recall.

Latent Action Models (LAMs) can capture manipulation priors from action-free videos, but discrete latents usually impose a granularity bottleneck, limiting their ability to represent fine-grained action. Recent continuous alternatives mitigate this, but they are typically used as proxies for world modeling and require a further planning step to produce executable actions. We thus introduce UniLAP, a unified continuous latent-action policy that models the latent space via flow matching and bypasses reliance on future frames during inference. By distilling a LAM teacher with a flow-adaptive objective, UniLAP inherits the teacher's action representation while retaining the multimodal distribution of feasible behaviors. To decouple latent representations from task-irrelevant appearance, we extract a label-free motion target to isolate relevant kinematic cues from spurious visual dynamics. With the established embodiment-agnostic latent manifold, UniLAP can be easily adapted to novel platforms. Pretrained on multi-source videos at a compact 1.1B parameter scale, UniLAP outperforms state-of-the-art latent-action and Vision-Language-Action (VLA) baselines across hand pose forecasting, single-arm and humanoid manipulation, and real-world deployment.

Physical AI is commonly formulated around an explicitly specified task: given a goal or instruction, a system perceives, reasons, and acts. In many real-world settings, however, important tasks emerge from continuously changing world states before a person explicitly formulates them. We study this complementary problem as Proactive Physical AI, where an AI system continuously observes the physical world, identifies situations that warrant intervention, and initiates appropriate actions at the right time. We develop ProAgent, an end-to-end system built around two components: a Jev-style proactive timing model (dubbed StreamingJev) that determines whether and when intervention is needed, and a Response Model that generates responses for people or invokes available device capabilities for physical execution. To support learning at scale, we develop an automated data pipeline that converts videos into supervision for proactive timing. Experiments demonstrate strong proactive performance under both explicit objectives and standing roles, combining more accurate intervention timing with fewer false triggers than the source-backbone baseline. Real-robot evaluations further demonstrate ProAgent's ability to identify emerging needs and translate them into appropriate physical actions.

Symbolic regression (SR) seeks explicit equations that explain observed data, and recent methods increasingly employ large language models (LLMs) and agents for equation discovery. However, these systems typically rely on manually designed workflows and prompts. We argue that the SR agent harness should itself be an optimization target: its design shapes search priors and how numerical feedback guides subsequent equation refinement. We therefore propose HarnessSR, a framework for jointly optimizing prompts and workflows to construct reusable SR agents. During both harness optimization and agent execution, HarnessSR supports either PySR or an LLM-based equation generator through tool-specific interfaces, allowing the framework to optimize search guidance for different equation-discovery mechanisms. HarnessSR first optimizes role prompts through equation-discovery feedback in basic workflows, then jointly searches prompt assignments and workflows with brief adaptation to align the two. Finally, it refines role instructions through reflection on residual patterns, candidate equation structures, and search outcomes, improving how these signals guide subsequent equation search. On LSR-Synth in LLM-SRBench, HarnessSR achieves a mean ID $\mathrm{Acc}_{0.01}$ of 61.33% across three backbones, a 45.0% relative improvement over the next-best baseline.

This article develops optimal order (up to logarithmic factors) weighted cubature rules for Monte Carlo numerical integration in Sobolev reproducing kernel Hilbert spaces over common matrix manifolds. Specifically, we consider matrix models of the Stiefel, Grassmann, flag, and Lagrangian Grassmann manifolds each over either real numbers $\mathbb{R}$, complex numbers $\mathbb{C}$, or quaternions $\mathbb{H}$. The proposed cubature rules use evaluation points which are independent and identically distributed uniform on the manifold. The weights are chosen to be optimal with respect to the reproducing kernel Hilbert space worst-case error, under the constraint that the weights sum to one. Crucially, this constraint ensures optimal weights may be found without having to evaluate the kernel mean embedding (the integral of the kernel), which is often difficult to compute. We consider computable Matérn kernels of Frobenius distances between matrices which reproduce the Sobolev reproducing kernel Hilbert spaces with equivalent norms. Open-source numerical experiments validate our theoretical development, including an application to simultaneous approximation of intrinsic volumes in the Cauchy–Kubota formula. We also present preliminary empirical results beyond independent points including transformations of low-discrepancy quasi-Monte Carlo points from the unit cube and jointly optimized points and weights.

Understanding how many aircraft, and which traffic configurations, can be safely and efficiently managed by air traffic controllers within a given airspace is fundamental to air traffic management. Air traffic controllers continuously modify aircraft trajectories through interventions such as speed advisories and radar vectoring, and the patterns of these interventions are difficult to capture effectively using existing air traffic simulators. To address this limitation, we propose Diffusion Transformers for Traffic Scene Generation (DiTSG), which learns to generate air traffic scenes from recorded multi-aircraft trajectories. DiTSG employs stacked multi-agent inverted Transformer layers as the denoising backbone of a diffusion transformer, jointly generates the trajectories of multiple aircraft within a traffic scene, and conditions the generation on agent-specific required times of arrival. The resulting trajectories generated by DiTSG preserve realistic inter-aircraft interactions observed in actual air traffic operations. Experiments show that the DiTSG generates diverse traffic scenes that reproduce the statistical properties of recorded traffic while preserving realistic inter-aircraft interactions.

Standard hate speech detection datasets often emphasize explicit harmful language, but directed harm can also be expressed indirectly, without clearly hateful wording. Sarcasm is one such form of indirect expression, as it can convey harmful intent through mock praise, exaggeration, humor, or apparent politeness. In this paper, we introduce SarcasTraps, a targeted external test set for analyzing the effect of sarcasm on Turkish hate speech detection. SarcasTraps contains 479 entries annotated along two dimensions: hate speech and sarcasm. We train classifiers using BERTurk and XLM-RoBERTa on two existing Turkish hate speech datasets and evaluate them on SarcasTraps, using it strictly as an external test set. We also evaluate GPT-4.1 and Llama-3.3-70B in a zero-shot setting on SarcasTraps. Results show that sarcasm substantially affects hate speech detection. Fine-tuned encoder models lose recall on sarcastic hate, while zero-shot LLMs maintain higher recall but suffer substantial precision drops on sarcastic examples. These findings show that both fine-tuned hate speech detectors and zero-shot LLMs struggle when sarcasm is involved, and SarcasTraps provides quantitative evidence of this challenge through a targeted Turkish test set. Due to the presence of sensitive and hateful language, SarcasTraps will be shared upon request for academic research purposes. Warning: This paper has examples from the dataset which contains offensive and hateful content towards groups of people.

Historical document analysis is crucial for cultural heritage preservation and low-resource language digitization. However, real historical documents often suffer from severe degradation, such as ink erosion and structural damage, causing mixed-visibility character sequences. Conventional OCR systems struggle in such evidence-limited scenarios, while contextual language models may over-correct visually clear characters. To address these challenges, we propose DDVR-Net, a degradation-aware decoupling and visibility-guided restoration framework for historical document recognition. DDVR-Net decouples intrinsic glyph structures from physical corruption at the feature level and predicts local visibility to estimate character reliability. A contextual decoder is selectively activated through a visibility-guided gate, restoring degraded characters while preserving clear visual recognition. Furthermore, we construct a low-resource multilingual benchmark covering Tibetan, Mongolian, and Uyghur scripts with 187000 document pairs, combining large-scale synthetic data and progressive training with real-world annotations to reduce the synthetic-to-real gap. Extensive experiments demonstrate that DDVR-Net consistently outperforms existing methods across three languages and diverse reading orders, providing an effective framework and benchmark for low-resource historical document recognition. Code and Benchmark: https://github.com/Shajiu/DDVR-Net

How do language models reconcile conflicting contextual evidence within a prompt? We study this question using lexical redefinition, where an explicit instruction specifies how a word should be interpreted and a subsequent sentence provides additional evidence through its usage. Across 870 word pairs, we find that lexical redefinitions are not applied as fixed rules. Instead, the behaviour depends systematically on the relationship between the explicit instruction and the intervening sentence: the contextualised usage can preserve, weaken, or completely reverse the instruction. These results suggest that the interpretation of an explicit instruction remains sensitive to subsequent contextual evidence. Lexical redefinition therefore provides a controlled behavioural paradigm for studying how language models integrate multiple sources of information within a prompt.

Convex relaxations make nonconvex optimization tractable by replacing hard nonconvex problems with convex surrogates that give valid bounds. Yet their design is still largely manual: a human expert must decide where to relax, which relaxation to use, and how much formulation growth is worth paying. We show that this design process can be learned. The method proposed in this paper, AutoConvexRelax, represents the symbolic formulation of a nonconvex problem as a tripartite graph linking variables, symbolic terms, and their objective or constraint positions. It then learns a policy that composes symbolic rewrites into problem-specific relaxation trajectories as the formulation evolves. We prove that finite action sequences generated by the learned policy preserve relaxation validity under explicit domain, occurrence-context, and representation-consistency conditions. We further establish computable error bounds for mixed relaxations and sufficient conditions for feasible recovery and global-optimality certification. As a first instantiation, we apply the method to nonconvex quadratic problems and fractional problems. On 241 benchmark quadratic instances, the learned policy strengthens McCormick lower bounds by 13.52%. Its mean lower bound is 3.70% below strengthened semidefinite relaxations on the normalized comparison metric, while it uses substantially smaller formulations. The learned policy retains this trade-off on fractional instances and reaches near-semidefinite bounds on beamforming and sensor-network localization instances. These results suggest that convex relaxation design can move from a fixed handcrafted choice to an automated learning process that adapts to problem structures. Code is available at https://github.com/auto4opt/AutoConvexRelax.

Descriptions of how people use AI are becoming training data for the next generation of models. We ask whether what a model reads about its users changes how it treats people. Using contrastive synthetic-document finetuning, we train Qwen3-32B on four matched corpora built from the same 1,625 real-world usage scenarios. The corpora differ only in the predominant motive they attribute to users of AI chatbots: helping others (prosocial), exploiting others (antisocial), legitimate self-advancement, or task completion. Finetuning shifts the models' stated beliefs about users, how true those beliefs look along an internal truth direction, and their broader judgments of how safe and good the world is. These social priors carry into behavior. In trust games, antisocial models expect less reciprocity and send less money than prosocial models. Their distrust is specific and targeted at people, and users of AI chatbots in particular. After an exploitative record, they send almost nothing to a person but most of their endowment to a random device with the same record. In a common-pool resource simulation, a single antisocial agent leads every group of otherwise sustainable agents to exhaust the resource, after privately planning to mislead them. Antisocial training also raises dark-triad scores and lowers alignment ratings. Belief shifts appear when antisocial documents are 1% of training tokens, persist when descriptions of assistant behavior are suppressed, and are not reproduced by training on discourse about misaligned AI. Capping Assistant-Axis drift and steering emotion-related directions leave the gaps largely intact. What models learn about the people they serve shapes how they subsequently treat them and others.

Recent advances in multimodal foundation models, exemplified by GPT-6 Astra, raise the prospect of general-purpose embodied reasoning and planning. Using Internet videos for embodied pretraining calls for reliable physical-scale annotations. Yet visual geometry foundation models often recover scene structure and camera motion without reliable metric scale. We show that uncertain size priors of familiar objects can jointly constrain a reconstruction's shared scale. Building on this insight, we present UniMetric, a universal and training-free metric scale recovery framework for geometry foundation models, without additional sensors or task-specific metric supervision. Specifically, Metric Prior Acquisition elicits object height intervals from a vision-language model. Semantic-Geometric Alignment measures reconstructed object heights along the estimated support-surface normal in each frame and converts the priors into scale intervals. Global Scale Recovery reconciles these intervals through a convex objective, yielding a jointly feasible scale when they overlap and a unique compromise otherwise. The estimated sequence-level scale is applied uniformly to reconstructed points and camera translations. Controlled experiments with real VLM priors show that combining observations improves scale accuracy with exact relative object heights. To evaluate metric accuracy on GFM reconstructions without ground-truth scale alignment, we introduce MetricBench, covering twelve embodied, driving, and roadside subsets with sensor-derived ground truth. Across nine geometry foundation models, UniMetric reduces mean depth Abs Rel, reconstruction Chamfer distance, and trajectory error by 54.2%, 83.8%, and 65.6%, respectively, improving every model on all three tasks at the suite level. We release the UniMetric code and MetricBench data.

We study the online fair division of indivisible goods. In the standard model, where assignments are irrevocable, no algorithm can guarantee any positive fraction of the maximin share (MMS). Reassigning earlier goods can restore fairness, but it may take value away from agents who already hold them. We ask how much fairness can be maintained if every reassignment must leave each affected agent at least as well off. We consider two forms of such upgrade recourse. Under item-level upgrades, each good an agent holds must be kept or replaced by a distinct good she values at least as much; under bundle-level upgrades, only her total value must not decrease. For item-level upgrades, we show that the optimal MMS factor is essentially $1/H_n \approx 1/\ln n$, where $H_n$ is the $n$-th harmonic number, and that fixed picking sequences nearly attain it. Since these sequences use only the agents' rankings of individual goods, knowing the actual values cannot improve the optimal factor for up to 33 agents, and improves it by at most a vanishing margin as $n$ grows. Bundle-level upgrades allow substantially stronger guarantees. By permitting several-for-one exchanges, they turn this vanishing factor into a constant: we achieve exact MMS for two agents and more than half of MMS for any number of agents, in both cases together with relaxations of envy-freeness. Finally, we show that predictions about future goods let online algorithms inherit any offline MMS guarantee, losing only an additive term bounded by the prediction error. Under item-level upgrades, however, insisting on exact MMS when predictions are correct leaves no positive guarantee when they are wrong.

Large audio-language models (LALMs) are fast becoming a general interface to sound, yet how far their broad pretraining carries into specialized domains remains largely open. Vessel monitoring, maritime safety, and ocean sensing all depend on recognizing underwater signals, but to our knowledge LALMs have not been systematically evaluated on them. Annotated recordings are costly to collect, and a faithful synthetic alternative has to bring together many interacting factors: the structure of detection pulses, communication waveforms, and ship-radiated noise, and the sound-speed profile, geometry, and multipath propagation through which each reaches the receiver. We address this by proposing UA-Bench, to our knowledge the first large-scale benchmark for evaluating LALMs on underwater acoustic signal recognition. It comprises 375,051 audio records that couple parameterized sources with modeled propagation, so that every record can be traced to the conditions that produced it. The records span these three signal families, and each carries a three-level label hierarchy (13 leaf classes), a templated multi-turn dialogue, and source and channel metadata. We evaluate on a balanced 2,600-record subset with a gated coarse-to-fine protocol that locates each error at source routing, fine-grained classification, or output format, and add a paired silent-audio control to test whether the waveform is actually used. Nine open LALMs, evaluated without task-specific adaptation, reach only 2.0%–13.1% leaf accuracy, against a random reference of 7.7%. The waveform helps source routing far more than leaf recognition (for one model, +29.5 versus +2.3 points over silence), and just four models gain reliably at the leaf level, by at most 4.7 points. UA-Bench thus gives a reference point for these models, and its released training partition supports follow-up adaptation work.

Nearest neighbor search algorithms are widely used and have a long history, but we find that key implementation details are seldom reviewed in articles about them. These details can materially impact performance, and thus change the conclusions one makes about a given algorithm's superiority. In this article, we highlight these impacts by reviewing the Cover Tree as an example, producing RustKNN, a unified Rust implementation of three construction variants of the Cover Tree with three query modes. Our analysis reveals: (1) the optimal query strategy depends on $k$ and on the dataset (single-tree search is fastest at $k{=}1$ on three of our four largest datasets, while dual-tree search is fastest at $k{\ge}5$ on the three largest), and choosing the wrong query mode can cost up to a $7.5\times$ factor in runtime; (2) the cover tree base parameter, set to 2 in theory and 1.3 in practice, is not a safe default for dual-tree search: 1.3 is optimal on only one of seven datasets, and on covtype it is $1.6\times$ slower than the optimum; (3) among five micro-optimizations, only early-exit distance computation has a consistent effect, and that effect grows with dimension (none at $d{=}32$, $1.31\times$ at $d{=}54$, and $1.49\times$ at $d{=}784$); and (4) all-nearest-neighbor and held-out benchmarks yield different implementation rankings. All main timing tables report medians over 20 independent allocations with bootstrap 95% confidence intervals. RustKNN is open source at https://github.com/CrowdStrike/RustKNN.

Tensor decoupling algorithms recover compact representations of multivariate functions by combining linear factor matrices with univariate internal nonlinearities. We identify two limitations of recent methods. First, their performance is not invariant to mismatched output or gradient magnitudes, and second, their spline internals require manual tuning of their degrees of freedom. We propose two lightweight fixes. The first is a preprocessing step that rescales the input Jacobian tensor, removing both the bias towards larger magnitude outputs and the ill-conditioning of certain factor updates caused by such a mismatch. The second replaces B-spline internals by penalized splines coupled with generalized cross-validation, making the method less dependent on the user selecting the right number of control points. Finally, we release an efficient open-source JAX implementation of our algorithm.

Dispersion requires a set of mobile agents to occupy distinct nodes of a graph such that each node contains at most one agent. We study deterministic dispersion of $k\le n$ agents on an $n$-node oriented square grid, where each node has globally consistent North, East, South, and West port labels, from arbitrary initial configurations. This includes both rooted configurations, where all agents are co-located initially, and general configurations, where agents are initially distributed arbitrarily across the nodes. On oriented square grids, Kshemkalyani, Molla, and Sharma~\cite{KMS20} gave an $O(\min(k,\Side))$-round algorithm in the local-communication model, where only agents on the same node can exchange information, and an $O(\sqrt{k})$-round algorithm in the global-communication model, where any two agents can communicate regardless of position, both using $O(\log k)$ bits per agent. The bound in the local communication model is optimal when $k=\Theta(n)$, but it remains open whether local communication alone can match the $\Omega(\sqrt{k})$ time lower bound when $k=o(n)$. We solve this by presenting Spread–Reverse–Collect (SRC), a deterministic algorithm that disperses any $k\le n$ agents in $O(\sqrt{k})$ synchronous rounds using $O(\log k)$ bits of persistent memory per agent, matching both time and memory lower bounds. SRC is based on a geometric primitive, a \sharpS-shaped parallel sweep, together with a simple priority rule. Unlike earlier grid algorithms that gather agents at the boundary and spread out from there, each SRC sweep starts from the agents' current position, and when it reaches the boundary the group simply reflects its sweep back inside and continues. SRC therefore needs no knowledge of the grid size and runs unchanged on an infinite oriented grid.

Online platforms often expose activity signals even when records of the underlying connections are incomplete or delayed. These activity responses may retain indirect evidence of connections that formed during the observation period but are absent from the recorded topology. Recovering such hidden structural changes can help keep network records aligned with evolving interactions and support downstream network analysis. However, the effects of a change propagate through the network and may overlap, making state changes difficult to associate with specific connections. We propose the Hierarchical Structural Context Ranker (HSCR), which combines visible network structure with node-state changes to infer connections that have formed but remain unrecorded. By jointly modeling the topological context of state responses and associations among potential connections, HSCR converts distributed activity changes into evidence for distinguishing potential connections without prespecified dynamical equations or parameters. Experiments on five temporal-network benchmarks and a real temporal Reddit network validate its effectiveness. Across observation spans and structural-record delays, true added connections achieve a 36%-50% reduction in mean true rank relative to the corresponding best baselines. This work provides a principled way to prioritize unrecorded additions for structural verification and record maintenance in online communities, information networks, and other networked systems.

Recent work has investigated Omni-modal Large Language Models (OLLMs), models which can take multiple modalities and their combinations as input, as a way forward to have systems really capable of interacting with their environment. Uncertainty quantification (UQ) for OLLMs, however, remains an open problem: existing methods largely target at most two modalities at a time, require access to model internals, or re-invocation of the evaluated model. We propose a model-agnostic UQ framework that scores a Chain-of-Thought response entirely through external omni-modal embeddings, measuring the grounding of the reasoning in the multimodal input and its internal coherence. Our contribution is the family of signals and an empirical characterisation of when it is reliable, rather than any single estimator: we find the geometric scores to be largely redundant with one another (mean within-split rank correlation 0.91-0.94) but strongly complementary to answer-dispersion baselines, and their reliability to depend on the generator and the modality. The resulting confidence estimates require no access to logits, hidden states or gradients of the evaluated model and no additional inference from it; they do require segmenting the reasoning trace with an auxiliary model and, for the aggregated score, a labelled calibration set that we show can be as small as 25 examples (although we also show that such a set buys little over fixing one signal a priori). On UNO-Bench, with both a closed and an open model, our scores yield small gains over self-consistency (systematic with the closed generator, mixed with the open one) and are competitive with self-verbalisation and internal-state baselines, especially in open-ended, omni-modal questions where previous state-of-the-art baselines struggle.

A digital twin is a virtual representation of a physical system, continuously updated with data from the physical system, enabling dynamic information exchange and decision-making. This bidirectional interaction, which distinguishes digital twins from conventional simulation models, introduces new challenges to simulation-based optimization, control, and learning. This paper reviews recent developments in decision-making methodologies motivated by digital twins. We focus on three interconnected areas: simulation optimization, sequential decision making, and risk-aware decision making. For simulation optimization, we discuss data-driven simulation optimization under input uncertainty, streaming input data, decision-dependent observations, and limited computational and data-collection resources. For sequential decision making, we review the connections among digital twins, stochastic optimal control, and reinforcement learning, with emphasis on Bayesian approaches that update models and policies as new information becomes available. For risk-aware decision making, we discuss distributionally robust and Bayesian risk formulations as alternative approaches for addressing model uncertainty. Throughout the paper, we emphasize the common problem of making decisions while the underlying model is being updated. We conclude by identifying research challenges and opportunities.

The expansion of the Internet of Things (IoT) in residential and local automation environments has highlighted the limitations of considering the internal network as an inherently trusted domain. Man-in-the-middle (MitM) attacks, such as ARP spoofing, can allow traffic interception or manipulation when communications lack appropriate confidentiality, integrity, and authentication mechanisms. This paper presents a systematized review of the security mechanisms used by two communication alternatives between ESP-family microcontrollers and the Home Assistant platform: MQTT protected through TLS 1.3 and the ESPHome Native API, whose encryption is based on the Noise Protocol Framework using the Noise_NNpsk0_25519_ChaChaPoly_SHA256 pattern. The analysis considers four dimensions: cryptographic robustness, attack surface, implementation complexity, and resource consumption. The review identifies significant differences resulting from the architecture and cryptographic model of each alternative. TLS 1.3 provides a standardized framework compatible with public-key infrastructure and formal identity management, though it introduces greater configuration complexity and computational overhead during session establishment. The ESPHome Native API employs a more specialized scheme with lower operational complexity, eliminates dynamic algorithm negotiation, and uses ChaCha20-Poly1305. The resource-consumption analysis relies on measurements published by the projects and manufacturers themselves, not on an experimental test bench, and this limitation is explicitly acknowledged. The results identify the conditions under which each mechanism is appropriate for local home-automation architectures and provide a basis for designing IoT systems that balance security, interoperability, and resource constraints.

Short-video search requires compact query rewriters that infer personalized intent from brief, ambiguous queries under strict latency constraints. The challenge is to decide when personalization is warranted and generate a reformulation that preserves the user's immediate information need. A rewriter optimized for fluency or engagement may still introduce unsupported details and redirect the search toward a different intent. Existing methods often underutilize users' actual reformulations and behavioral traces, relying instead on synthetic reasoning and indirect behavioral rewards that may produce more engaging candidates but cause intent drift. To build an effective rewriter that remains faithful to user intent, we propose WeFair, Fidelity-Aware Alignment from Human Feedback for Personalized Query Rewriting in WeChat short-video search. WeFair assigns real user behavioral signals to a progression from demonstration to preference to verification. We first mine actual human reformulations from search sessions, then apply multidimensional LLM screening and contextual verification to filter unsupported transitions without extensive manual annotation. During training, we begin with a supervised curriculum that progressively builds rewriting capabilities from large-scale human personalized query demonstrations, and then use direct preference optimization to align with users' subsequent consumption. In the final on-policy optimization stage, we construct a strict verification reward based on actual human reformulations to further boost performance with group relative policy optimization. Further analysis shows that auxiliary rewards can activate more rollout groups in GRPO but reduce intent fidelity. Online A/B testing in WeChat short-video search shows a 1.576% relative increase in effective clicked-query rate (VV>10s) and a 1.731% relative reduction in query reformulation rate.

We study the computational complexity of graph distortion distances, a family of graph metrics measuring how vertex distances differ between two graphs. Formally, these distances are defined in terms of standard matrix norms on the difference of distance matrices under an optimal vertex alignment. The resulting metrics have natural interpretations, corresponding to three ways of measuring the distortion of shortest-path distances: the maximum distortion among all vertex pairs (entrywise maximum norm), global aggregation (entrywise $p$-norms), and local aggregation (operator $p$-norms) of distortions. For maximum distortion, we show that approximation below a factor of two is NP-hard even for highly restricted trees. Due to the close connection to the Gromov-Hausdorff distance, we obtain a similar hardness result for the latter. In both cases, we show essentially optimal additive inapproximability. On the algorithmic side, we give a subcubic-time algorithm for optimal embeddings of connected graphs into paths, based on a structural decomposition along a diametral path. Additionally, we give a simple approximation algorithm when one input graph has a "relatively path-like" metric structure, i.e. has bounded pathlength. For local and global distortion, we derive various hardness results even for restricted inputs. Making use of a close connection between the global distortion distances and Gromov-Wasserstein distances, we give the first formal proof of NP-hardness of the latter, which has been claimed in the literature many times but has, to the best of our knowledge, never been correctly proven.

Keeling et al. [Keeling2026Generative] report on a case study recently conducted by Redgrave LLP in which they assert that Relativity's Generative AI Review tool (aiR) achieved 88% recall while Relativity Assisted Learning (RAL), its machine-learning tool, achieved 64% recall. These results are not comparable because the methods employed have different objectives, because of inter-assessor disagreement, and because they represent different recall-precision and cost-benefit trade-offs. Our analysis of the Redgrave data indicates that, when these differences are taken into account, aiR and manual review are comparable. Furthermore, established machine-learning models employing Continuous Active Learning (CAL) show state-of-the-art effectiveness, and superior efficiency, when trained only on a sample of 2,000 first-pass-reviewer responsiveness determinations.

Although spiking neural networks (SNNs) show promise in energy-efficient computing through event-driven computation and stateful temporal processing, their recurrent membrane dynamics render their training inherently sequential and inefficient for long sequences. Existing methods achieve full-sequence parallelism in SNNs through two main strategies: redesigning or approximating spiking dynamics to avoid reset-coupled temporal dependencies, or relying on iterative relaxed fixed-point solvers that suffer from excessive memory costs and iteration-dependent backward overhead. To address these limitations, we present FidLIF, a novel framework that reconciles the incremental update and hard-reset behavior of standard spiking LIF neurons with parallel forward propagation and BPTT-equivalent backward propagation. FidLIF iteratively recovers the full-sequence membrane-potential trajectory in parallel. By exploiting the lower bidiagonal structure induced by the local recurrence of LIF neurons, FidLIF formulates each forward trajectory correction and backward adjoint propagation as associative scans with an $O(\log T)$ temporal depth. This formulation also supports input-dependent neuronal parameters and allows the same neuron to perform full-sequence parallel training and incremental step-by-step inference at deployment. In our numerical convergence evaluation, FidLIF achieves a mean absolute error below $10^{-7}$ relative to the matched serial relaxed LIF dynamics, while yielding gradients numerically equivalent to exact BPTT. Our Triton implementation delivers favorable long-sequence latency and memory scaling, while achieving competitive performance on four sequential classification benchmarks. By reconciling parallel training with step-by-step execution, FidLIF enables high-fidelity spiking dynamics to benefit from parallel hardware without compromising incremental inference at deployment.

We develop a comprehensive theoretical analysis of Matrix Shampoo for tracking time-varying optima in nonstationary stochastic optimization. We derive finite-horizon high-probability bounds for average tracking error that disentangle initialization, optimizer drift, momentum-induced drift penalty, stochastic noise, and adaptive preconditioning, with explicit dependence on momentum parameter \(\beta_1\) and stabilization parameter \(\varepsilon\). These bounds reveal an explicit drift–noise and stability–geometry tradeoffs: noise-dominated regimes favor larger $\beta_1$, which provides stronger first-moment averaging and greater suppression of gradient noise, whereas drift-dominated regimes favor smaller $\beta_1$, allowing faster adaptation to moving optimizer. Larger $\varepsilon$ reduces the preconditioner-perturbation term while weakening Shampoo's ability to exploit matrix geometry, thereby reducing adaptivity. We establish a matching minimax lower bound under squared path-variation constraint, proving the information-theoretic optimality of Shampoo over the class of causal optimization procedures. Under a constant stepsize, we establish high-probability bounds for the average projected stationarity gap along general nonstationary nonconvex objectives. We validate our theoretical findings on both synthetic and multiple real-world problems.

Online motor imagery-based brain-computer interfaces (MI-BCIs) typically decode continuous EEG using fixed-length sliding windows that may be misaligned with neural events. Existing window optimization methods often evaluate windows independently, overlooking temporal dependencies and potentially producing fragmented or physiologically implausible selections. We propose Temporal Structure Embedded Reinforcement Learning (TSE-RL), a framework that incorporates prior knowledge of MI event structure into a constrained Markov decision process for adaptive window optimization. First, TSE-RL assigns each window an explicit semantic role through Discard, Onset, Sustain, or Offset actions, extending binary retain/discard decisions to event-level alignment. Second, a finite-state machine masks invalid actions to enforce the temporal order of MI phases and promote physiologically plausible window sequences. Third, a Transformer-based history encoder captures dependencies across windows, improving training stability and computational efficiency compared with recurrent architectures. TSE-RL is plug-and-play and can be integrated with diverse MI classifiers. Experiments on multiple benchmark datasets demonstrate consistent improvements in decoding performance in both cross-session and cross-subject settings. Visualization analyses further suggest that the learned window selection policy aligns with neurophysiological principles.

Antibody therapeutics have transformed treatment across diverse therapeutic areas and have become central to modern medicine. While recent machine learning approaches aim to accelerate antibody discovery through structure-based design and target-affinity optimization, successful candidates must also satisfy broad developability constraints, including structural compatibility, low immunogenicity, and favorable biophysical properties, while remaining sufficiently diverse to provide sufficient coverage of design space to mitigate later-stage development failures. We introduce a preference-conditioned multi-objective generative flow network (GFN) fine-tuning pipeline for fixed-framework CDRH3 loop redesign that fine-tunes a biological sequence model against many inexpensive proxies covering backbone compatibility, sequence likelihood, immunogenicity risk, novelty, and developability liabilities. It combines up to fourteen predictive rewards in a single policy trained to sample CDRH3 loops across 623 SAbDab antibody structures, leveraging a reward reduction strategy to preserve Pareto dominance under this learned map. We also introduce a novel autoregressive span-in-filling variant of AMPLIFY-350M that serves both as a scaffold-conditioned CDRH3 generation prior and as a direct conditional likelihood scorer. We evaluated reward scores, diversity, and design quality on loops sampled on 45 held-out antibody complexes to assess structural consistency, confidence, aggregation propensity, stability, and antigen binding. Compared with the inverse-folding methods ProteinMPNN and ESM-IF, our approach produces roughly 10x more distinct acceptable candidates per design at 50% sequence identity, while holding 10% of the 14-objective Pareto front pooled across methods, trading per-sample quality for greater exploration of novel possible structures.

Machine learning engineering (MLE) agents are emerging as a new paradigm for automating end-to-end ML pipeline development. Using large language models for code generation and predictive performance as a machine-gradable feedback signal, these agents iteratively synthesize, execute, and refine pipeline code. Yet current systems are largely evaluated on pre-built, single-table benchmarks that abstract away key data management challenges. This tutorial introduces the foundations of MLE agents and highlights research opportunities for the database community. We cover how agents combine program synthesis with iterative and evolutionary optimization, and discuss open challenges in efficient pipeline execution, data discovery and augmentation over heterogeneous sources, reliability and debugging, as well as fairness and compliance. We connect these challenges to existing data management techniques and conclude with a hands-on session using contemporary MLE agents.

The rapid expansion of artificial intelligence (AI) is driving substantial growth in data center electricity demand. Interval-based energy blocks in day-ahead markets do not explicitly retain predictable intra-interval demand evolution, potentially creating mismatches with ramp-feasible generation trajectories and affecting settlement. This paper proposes a power-trajectory framework in which data centers declare predictable aggregate demand evolution through continuous demand-side power trajectories before market clearing. The declared trajectories are treated as exogenous inputs for constructing generation trajectories and the corresponding price trajectory. For feasible market-clearing outcomes, physical temporal consistency is established through continuous-time power balance and satisfaction of the modeled generation operating constraints, while economic temporal consistency is established by preserving the dispatch-price correspondence in energy settlement on a common temporal basis. Under the adopted settlement definitions, a trajectory-based settlement analysis separately quantifies the monetary magnitude of representation-induced power imbalances and the supply-side compensation that may remain under temporally consistent clearing and settlement. An illustrative 5-bus case demonstrates the elimination of representation-induced power imbalance and identifies the timing and physical sources of supply-side compensation. A modified IEEE RTS-GMLC case further illustrates these compensation effects across multiple generating units under workload-driven data center demand variations.