Papers

Active IP geolocation techniques rely on the responsiveness of Internet hosts, while passive techniques depend on data sources that are unevenly adopted and prone to staleness and error. In this work, we invert the active IP geolocation problem by listening at a geographically distributed set of vantage points for unsolicited Internet scans. By reactively completing connections with these scanners, we obtain round-trip time (RTT) measurements from hosts that may otherwise be unresponsive and use those measurements for geolocation. Over the course of 20 days in August 2026, we recorded 46.8 million scans from 289,746 distinct scanners, with a reactive RTT for each scan. We show that reactive RTTs align with those obtained by sending active probes, with more than three-quarters of 990,204 measurement pairs differing by at most 10 ms. Over half of the scanners that yielded a reactive RTT did not answer our active probes, highlighting the value of our approach. We demonstrate the feasibility of multilateration using purely reactive RTTs for 56,112 distinct IP addresses. Finally, using speed-of-light constraints, we refute commercial geolocation claims for 6,852 addresses.

Real-world deployment of active learning typically forces practitioners to choose an acquisition strategy before any data is labeled. This is a daunting task: strategy performance varies widely across settings (e.g. datasets, surrogate models) and cannot be assessed without deployment. Existing strategy selection methods explore one strategy from a portfolio at each round and identify the optimal one using bandit feedback or model retraining. Many acquisition rounds are therefore spent exploring strategies rather than collecting the most informative data. Such overhead is a major barrier to AL-driven design of high-throughput experiments, such as genetic perturbation screens and directed evolution, where AL runs consist of only a few rounds with large batch sizes. This regime permits a natural alternative: acquiring data using multiple AL strategies within a single batch. We refer to this as soft strategy selection and introduce FractAL, a method specifically designed for this task. FractAL infers a per-strategy reward using influence-function-based data attribution, which requires no additional retraining, and then computes budget shares for each strategy in the portfolio using online mirror descent. We benchmark FractAL across 7 setups spanning classification, regression, and genetic perturbation effect prediction. The results highlight that strategy selection is a hard problem: every existing method performs worse than random sampling on at least one setup. FractAL, however, matches or outperforms every baseline, including random sampling, on all 7 setups. Its allocations concentrate budget on the strongest strategies in the portfolio while pruning the weakest. FractAL is therefore a reliable choice for real-world deployments, where the optimal strategy is unknown in advance, an important step towards making AL practical for high-throughput experiments and modern scientific discovery.

Control barrier functions are an effective model-based tool to formally certify the safety of a system. However, transferring their theoretical guarantees to real-world robotics systems requires high model fidelity. For example, payloads or wind disturbances can cause significant model perturbations to an aerial vehicle, leading to safety compromises. In this work, we propose a certifiable online learning-enhanced robust adaptive control barrier function, which adapts to disturbances using a Neural ODE and quantifies its adaptation uncertainty with conformal prediction. Our approach guarantees safety at all time under unknown time-varying model disturbances. It adopts a conservative strategy when the adaptation uncertainty is high; and efficiently adapts to reduce controller conservativeness as it receives more data. Our approach provides a provable safety guarantee with a probability bound under suitable Lipschitz smoothness assumptions on the underlying model and trajectory. These results demonstrate the potential of our method as a practical safety controller for robotics system operating under model perturbations.

Mechanistic interpretability has emerged as the primary means to understand safety behavior of LLMs. However, existing tools primarily focus on the activating neurons or features of a model. The role of the remaining large set of inactive components is invisible to such methods. This work demonstrates that the inactive set contains safety-critical features that are causally relevant for refusal of harmful prompts. Suppressing such features could turn refusals into compliance, while passing undetected by prevalent interpretability tools. We introduce the Counterfactual Activation Potential (CAP), a metric that quantifies a suppressed feature's latent activation tendency as the product of its encoder alignment (how strongly the input drives it), suppression strength (how strongly active features inhibit it), and safety criticality (how much refusal depends on it). To find suppressed safety features at scale, we propose CAP-guided Safety Feature Discovery (CSFD), a two-stage filtering algorithm that identifies candidate safety features from hundreds of thousands of transcoder features without exhaustive ablation. A significant fraction of trials turn compliant with harmful prompts when a candidate feature is ablated. Under natural jailbreaks, the suppression acting on the highest-CAP features rises 2-4x, and their activation correspondingly falls by up to 80%. Amplifying a feature's suppressors pushes its activation down and raises harmful compliance with prompts related to the suppressed feature, with no such effect for random features. Our experiments span five Gemma, Qwen, and Llama models across various parameter sizes. Our findings indicate that jailbreaks could operate in part by suppressing safety-critical features rather than solely activating harmful ones, and that suppressed features are a necessary complement to activation-focused interpretability of safety behavior.

Natural language to SQL (NL-to-SQL) benchmarks are foundational to progress in data analysis research, yet recent work has shown that widely-used benchmarks contain significant annotation errors. These errors silently corrupt evaluation metrics, penalize correct model output, and distort the field's understanding of state-of-the-art performance. We present SALUS, a system that automatically detects annotation errors in NL-to-SQL benchmarks. SALUS frames benchmark auditing as a weakly supervised error detection: SQL generated by multiple LLM agents drive a suite of complementary weak-labeling functions. By passing this noisy vote matrix through a generative label model, we extract high-confidence training samples without requiring human ground truth. These samples train a decision plane that maps gold SQL query features to per-agent trustworthiness, allowing SALUS to intelligently fuse reliability estimates with raw verdicts for rigorous benchmark error detection. We evaluate on BIRD-Clean-xs, a benchmark of 298 BIRD development tasks with manually verified correctness labels. SALUS achieves F1 = 0.9194, significantly outperforming the state-of-the-art baselines. Applying SALUS to the full development sets, we estimate annotation error rates of approximately 37% on BIRD and 27% on Spider.

We introduce Loop Flow Transformers (LiFT), a family of looped generative models that scales computation by repeatedly applying a shared Diffusion Transformer (DiT) core, with only light changes to the standard architecture. Rather than asking every recurrent step for the final prediction, LiFT trains each step with a single regression target: a point on a straight path from the model's initial estimate to the flow-matching target. Because we index these targets by a continuous depth coordinate, a trained model can loop far beyond its training depth with no retraining, early exits, or other modifications. In our experiments, these longer rollouts improve generation, so inference computation can grow without adding parameters. On ImageNet at 256x256, LiFT-L/2 achieves an FID 3.34 points lower than our dense DiT-XL/2 baseline while using approximately 60% fewer parameters, 32% fewer training FLOPs, and 52% fewer inference FLOPs.

Human–LLM interaction datasets shape our understanding of AI use and provide a foundation for downstream research, including training and evaluation of user models. In recent years, a growing number of datasets have sought to capture a representative picture of human–LLM interactions. But how different are the pictures these datasets provide, and what do those differences mean for research built on them? We study these questions across seven conversation datasets, spanning in-the-wild chat logs and human preference data. We begin by revisiting the dataset classification experiment of Torralba & Efros and find that neural network classifiers identify the source of a conversation from user messages alone well above chance, indicating distinctive dataset signatures. This separability persists after matching datasets on the dimensions of human-designed taxonomies, implying subtle differences that these taxonomies do not capture. We then examine the implications for user modeling: how dataset signatures propagate to the outputs of user models trained on these datasets; how dataset choice influences evaluations of user model quality and subsequent evaluations of LLM assistants paired with these user models; and how dataset classifiers can guide data selection for training user models. While each dataset is meant to capture a slice of 'real-world' interactions, our findings reveal the extent to which these slices diverge, and the consequences of those differences for research built on these foundations.

Harness search keeps a change to the prompts, reasoning switches, token budgets or parsers around a frozen model if the change raises a score. Such a gain can come from answers the parser could not read before, a weak comparison, or a defect in the evaluation. We preregistered a study of where these gains come from, with three small models, three benchmarks, replication and test partitions, a GEPA search arm and six evaluation defects injected one at a time, and we report all 47 primary endpoints. Turning thinking off raised accuracy over a capped thinking setting in 5 of 9 model-benchmark cells, and in each the gain came mostly from questions where the capped setting gave no readable answer. The thinking-off setting was not meaningfully worse than a rescue configuration or four GEPA-selected harnesses in 11 of 13 comparisons, and lost to the rescue on GSM8K for two models. GEPA repaired its broken starting points, but none of its selected harnesses was more accurate than the thinking-off setting. A thinking budget in the serving engine, which also allows a longer answer, lowered truncation and raised the parse rate in 6 of 9 cells. In 6 of 15 evaluable defect-model pairs, replication through the same pipeline reproduced the defect's distortion instead of revealing it. On the LongevityBench multiple-choice tasks, only the longevity-tuned model beat the strongest constant-label baseline.

AI agents that send emails, edit files, and make purchases must decide when to act on their own and when to check with the user first. This decision is usually evaluated by showing a model a proposed action, asking whether it should proceed, and scoring agreement with human labels. We introduce DelegationBench to test whether such scores can be trusted. It has 156 scenarios with four possible responses (act, ask for permission, ask for missing information, refuse), and most scenarios come in matched pairs that change a single feature: whether the action was requested, what is at stake, whether it can be undone, or who will see it. Across ten models from five families, agreement scores mislead in three ways. A simple keyword rule, which we wrote after seeing the benchmark, agrees with our annotators more often than eight of the models, yet its decision changes in only 9 of 48 matched pairs. Equivalent ways of asking the same question change how often a model acts by up to 52.5 percentage points. And every model stops to ask the user less often when it must carry out the task with tools than when it judges a proposed action. When rules are stated explicitly, the same models follow them almost perfectly, so the gaps are not explained by a general inability to follow rules. We release the benchmark and evaluation tools and recommend reporting these properties separately rather than as one score.

Embodied systems need geometric perception that exploits available observations beyond images alone. Recent feed-forward 3D models incorporate geometric priors, including camera poses, intrinsics, and depth. However, handling noisy poses, preserving accurate priors, and recovering physical scale require more than simply accepting these inputs. We introduce Vision-Prior Geometry Grounded Transformer (VPGGT), a VGGT-based framework that extends OmniVGGT for prior-aware embodied perception. We formulate sensor-motivated pose corruptions from ground-truth trajectories for training and introduce a parameter-free prior residual connection (PRC) to mitigate prior dilution, where predictions are less accurate than their supplied pose priors. Our noise formulation targets camera poses; supplied intrinsics and depth receive no additional corruption. We further introduce Metric Global Attention, which conditions a global scale token on available pose and depth scales and predicts a shared metric scaling factor for the geometric outputs. Experiments across four datasets show that PRC improves translation-direction accuracy and joint pose AUC over a matched training baseline when camera priors are provided for all views, under both exact and corrupted poses. These results support explicit prior access during refinement as a useful addition to feature-level conditioning.

We formulate a novel optimal control problem for data-driven dynamic resource allocation in multicore real-time systems. The proposed formulation accounts for hard deadline constraints and resource bounds. Under mild assumptions, we prove several results on the feasibility and existence of optimal solution for the proposed formulation. Building on these analyses, we design an algorithm to solve the multi-core dynamic resource allocation problem. We demonstrate the practical use of this new algorithm by experimentally evaluating it on a real multicore platform using four benchmarks from the literature.

We present DynaMesh, a dynamic texture generation method for 3D meshes. Given a textureless shape and a text prompt describing an effect, our method produces an appearance that evolves while the object's geometry remains unchanged. Previous works on dynamic 3D content generation have focused on motion, where an object's geometry and position change while keeping its appearance the same. Methods on texture generation sit on the other side of the problem, painting appearance onto a shape as a fixed surface property and not as an evolving process. Neither addresses a visual effect that propagates on a 3D object. A natural route consists of two generators: a video model that shows the effect from a single view, and an image-to-3D generator that lifts each frame to 3D. However, the latter has no notion of time, so running it per video frame produces a sequence that flickers, loses effect details, and yields a different mesh at every video frame. Our method addresses these failures by conditioning a video model on a render of the mesh and the prompt to obtain a reference video, then running a frozen image-to-3D generator on the video with two changes. The conditioning of each frame is blended over a temporal window, and low-rank adapters are fit per shape to restore the lost details. The mesh is encoded once for the whole sequence, so geometry is constant by construction, and the output is a single mesh with a texture per frame. Applied to various objects and effects, DynaMesh substantially improves over recent video-to-4D and texturing methods, and can generalize its temporal effect to different shapes never seen during training. Our project page is at https://threedle.github.io/dynamesh/.

Egocentric assistants must connect what users say with what they see across long interaction histories. We formalize this challenge as Spatially grounded Conversational Reasoning (SpaCR): cross-scene, recall-oriented, and counterfactual spatial queries that combine user-stated facts with geometric evidence. Direct vision-language models incur high inference costs and context limits as histories grow, while keyframe selection and retrieval can omit objects or evidence needed for complete recall. We propose Spatially grounded Conversational Memory (SpaC-MEM), an object-centric working memory that uses 3D reconstruction and segmentation to ground conversational information in persistent physical objects. It compresses multimodal histories while preserving spatial evidence and allowing object-specific facts to be updated through dialogue. We also introduce Ego-SpaCR, a benchmark comprising 620 ScanNet video sessions augmented with 95 task-oriented conversations and 3,100 evaluation queries. SpaC-MEM achieves the highest overall answer accuracy among the evaluated methods and improves object recall while requiring substantially fewer reference input tokens than native-video baselines. Removing 3D spatial information substantially degrades performance, highlighting the importance of preserving spatial and conversational evidence together.

Variational quantum algorithms turn choices of Hamiltonian, ansatz, and parameterization into a classical nonconvex optimization problem. We study how this objective function can be visualized and characterized in ways that help explain optimizer behavior. We distinguish two properties of the objective: the number of local minima encountered along sampled directions and the differences in quality among local-search endpoints. We then ask how a local optimizer, represented by BFGS, compares with adaptive differential evolution, represented by jSO. Rather than comparing jSO with a single local run, we allow BFGS multiple starts within the same function-evaluation budget. This gives local search repeated opportunities to explore different basins and provides a stronger baseline for asking when a global evolutionary solver is useful. We study these questions using VQE and QAOA, focusing on how frustra- tion, circuit depth, parameter tying, mixed locality, and nonlinear repa- rameterization change the Hamiltonian expectation-value objective seen by the classical optimizer. Increasing independent circuit depth raises the sampled local-minimum count without making global search more effective. Parameter tying, by contrast, produces both more repeated local structure and much larger differences in quality among local-search endpoints, and in this regime adaptive differential evolution outperforms function-evaluation-matched multistart BFGS. The comparison shows that the number of local minima alone does not determine whether global search is advantageous: the important distinction is whether different basins lead to similarly good solutions or to substantially different objective values. These results provide a practical workflow for connecting model construction, objective- function geometry, empirical diagnostics, and optimizer choice.

This letter develops a physics-informed one-shot beam tracking method for stacked intelligent metasurface (SIM)-enabled low-Earth-orbit (LEO) links with $K$ users and an $L$-layer SIM with $N$ meta-atoms per layer. The predictable orbital motion is propagated through the line-of-sight (LoS) geometry to forecast the finite-interval variation of the user signal-to-noise ratio (SNR) profile, and an analytical multilayer Jacobian relates this variation to the $LN$ SIM phase coefficients. Beam tracking is formulated as normalized SNR-profile preservation and solved through a single regularized Gauss–Newton correction that requires one Jacobian evaluation and a $K\times K$ linear solve, thereby avoiding iterative reoptimization. A tracking-error bound separates the regularization residual from the nonlinear Taylor remainder and establishes a second-order local error under a full-row-rank Jacobian. Numerical results show that the proposed method retains approximately $94\%$ and $92\%$ of the iterative-reoptimization rate for $L=2$ and $L=4$, respectively, while running about $350\times$ and $260\times$ faster at $N=225$, and confirm the predicted second-order residual error behavior.

The Fréchet distance is a de facto standard for evaluating generative models across domains, appearing as FID for images and FVD for videos. It summarizes the discrepancy between generated and reference distributions in a single scalar, with lower values typically interpreted as better generation quality. However, this scalar view can obscure what drives the comparison. For example, in COCO dataset, increasing the number of diffusion sampling steps improves ImageReward scores yet worsens (increases) FID. Motivated by this mismatch, we seek to make the Fréchet distance more interpretable by uncovering where the discrepancy lies. To this end, we introduce directional Fréchet distance, the expected squared projection of the optimal transport displacement onto a given direction. Across our image, video, and protein case studies, we find that a small number of interpretable directions account for much of the distance. We use these directions to explain the FID increase in terms of semantic concepts represented by CLIP embeddings, quantify FVD's bias toward per-frame appearance, and revisit the interpretation of Protein FID. We open-source our codebase at https://github.com/yhlee-add/directional-fd.

We study the problem of computing the tightest \(\alpha\)-potential approximation of a general-sum game over continuous action spaces, within a prescribed class of potential functions and when each player has access only to its own utility function. The difficulty is twofold: the approximation error involves a worst-case search over an infinite set of unilateral deviations, and the required utility information is distributed across players. For a linear-in-parameters potential class, we use an exact finite-tuple reformulation that separates the problem into a global outer search over deviation tuples and distributed convex inner problems. We develop a primal–dual inner oracle tailored to this structure and establish a uniform one-sided accuracy guarantee. This oracle can be combined with global outer search to obtain an end-to-end guarantee on the outer optimization error. We also develop a projected zeroth-order outer method as a computationally lighter alternative for higher-dimensional problems. Numerical experiments illustrate the accuracy–computation tradeoff between the two outer-search methods and show that the proposed optimization framework can improve upon analytical \(\alpha\)-potential constructions.

A new analytical technique is proposed for calculating adequacy metrics for generation planning. The proposed technique is simple, direct, and provides higher accuracy than traditional discretized approaches. Additionally, a new analytical methodology is introduced for calculating loss of load days (LOL Days) as distinct from loss of load events or frequency of load loss. This paper emphasizes that, where feasible, analytical methods are preferable to Monte Carlo simulations because they yield exact closed-form solutions rather than statistical estimates.

We develop and analyze several families of fully discrete, high-order methods in the finite difference time domain (FDTD) framework for a nonlinear dispersive electromagnetic model for ultrashort pulse propagation in a nonlinear optical medium. The model consists of the time domain Maxwell's equations coupled to two ordinary differential equations; the nonlinear Kerr-Debye equation and the linear dispersive Lorentz equation, in one space dimension. The proposed methods combine staggered spatial discretizations of $2M$th-order, for integer $M>0$, with second-order explicit leapfrog or implicit trapezoidal time integrators. To handle the stiff relaxation in the nonlinear Kerr-Debye equation, we adopt a modified exponential integrator, introduced in \cite{PengJCP2020}, which ensures asymptotic-preserving (AP) behavior in the limit of vanishing relaxation time. We prove that the resulting schemes satisfy a discrete analogue of the continuous energy identity and preserve the positivity (PP) of the nonlinear susceptibility. For the schemes based on the leapfrog time integrator, we establish energy stability under a CFL condition, while the schemes based on the trapezoidal time integrator are unconditionally energy stable. High-order temporal accuracy in the trapezoidal time integrator based schemes is further achieved via Richardson extrapolation yielding $2M$th-order accurate methods in both space and time. Numerical experiments demonstrate the theoretical convergence rates and the ability of the proposed schemes to capture nonlinear wave phenomena such as soliton propagation, self-steepening, and odd harmonic generation. This work extends previous AP-PP and energy stable approaches in a discontinuous Galerkin framework to the FDTD setting, offering a computationally efficient, flexible and robust tool for simulating electromagnetic wave propagation in nonlinear dispersive optical media.

Autonomous Mobile Robots (AMRs) on active construction sites face severe navigational challenges: geometry-based traversability mapping (e.g., LiDAR) misses visually hazardous but geometrically flat surfaces like wet mud and ponding concrete, while abrupt geometry on drivable speed-breakers and inclines produces phantom obstacles. We propose a real-time, failure-mode-aware multimodal traversability pipeline on an NVIDIA Jetson AGX Orin, where LiDAR is the primary geometric safety estimate and monocular semantics act as a selective, class- and confidence-gated corrective signal. The representation retains distinct traversable classes, namely flat road, terrain, and rocky terrain, while flagging construction hazards. We also release a multimodal construction-site dataset from a custom AMR: four closed-loop ROS 2 sequences from two active sites (RGB, depth, LiDAR, IMU, GPS-RTK, odometry) plus 506 annotated frames across 28 semantic classes. By projecting LiDAR onto dense semantic masks, resolving sparsity via morphological in-painting, and applying failure-mode-aware fusion with Patchwork++, the system corrects complementary geometric failure modes for a local AMR costmap.

Different low-rank compression methods can produce compressed LLMs that respond differently to the same post-compression recovery procedure, and relative advantages observed between methods at the compression endpoint may shrink, grow, or even reverse after recovery. We ask whether this recovery heterogeneity reflects functional structure beyond scalar loss evolution, and how that structure evolves throughout recovery. Our results establish that this heterogeneity reflects a reproducible compression-induced functional structure, which we formalize as recovery pressure. To characterize this structure consistently throughout recovery, we develop a standardized functional characterization within each backbone that is applicable across heterogeneous low-rank methods. The primary backward characterization reveals reproducible module-wise structure across independent probes, while a complementary forward-only characterization recovers related structure without loss or backpropagation. We further find that recovery pressure measured at the endpoint is associated with subsequent recovery response; during recovery, its module-wise structure is reorganized non-uniformly, and localized updates induce distributed responses beyond directly updated modules. Further evidence indicates that tracking this evolving structure provides a complementary functional view of recovery progress alongside scalar loss.

Large-scale EEG foundation models have demonstrated promising transferability across neurological disorders, but often require millions of parameters and substantial computational resources. In this paper, we present the Universal Semantic EEG Foundation Model (USE-FM), a lightweight EEG foundation model that learns transferable neural representations through self-supervised signal reconstruction on the Temple University Hospital EEG Corpus (TUEG). After pretraining, the encoder is frozen and evaluated on two clinically distinct downstream tasks, abnormal EEG detection (TUAB) and epileptic seizure recognition (TUEP), using a unified frozen-transfer protocol against recent EEG foundation models, including LUNA-Base and CBraMod. With only 1.46 million parameters, approximately one-fifth the size of existing models, USE-FM achieves competitive overall performance, including strong sensitivity and F1-score on TUEP (SEN $75.00 \pm 14.14$, F1 $70.37 \pm 4.01$), while maintaining competitive performance on TUAB (AUC $85.24 \pm 5.61$). Beyond downstream classification, latent representation analysis using $k$-means clustering together with PCA and t-SNE demonstrates that USE-FM learns organized semantic EEG representations comparable to substantially larger foundation models. These results suggest that large-scale self-supervised pretraining enables lightweight architectures to learn transferable semantic EEG representations, providing a computationally efficient foundation for cross-disorder analysis and future clinical decision support in neurological disorders.

The rapid growth of AI data centers is creating concentrated electricity demands that can exceed available distribution network headroom and delay interconnection. This paper investigates whether capacity can be used effectively by coordinating flexibility on both sides of the interconnection. A day-ahead mixed integer second order cone programming framework maximizes feasible AI data center IT capacity using utility side conservation voltage reduction and network topology reconfiguration, together with data center workload shifting and reserve constrained uninterruptible power supply storage. Background feeder demand uses voltage dependent ZIP models, while the data center is modeled as constant power demand using Training, Inference, and Mixed workload profiles. A two-stage solution first maximizes interconnection capacity and then minimizes feeder losses for the retained capacity, with numerical relaxation screening. Studies on a 116-bus model derived from the IEEE 123-node feeder show that, at the constrained bus-60 connection point, grid side flexibility, driven almost entirely by network reconfiguration, increases feasible IT capacity by approximately 42.5-45.0%, while data center flexibility alone provides approximately 2.1-5.4% gains. Coordinated operation increases feasible IT capacity from approximately 10.5-10.6 MW under BASE operation to 15.3-16.0 MW, corresponding to gains of approximately 45.9-50.8%; the 16.0 MW Inference result is limited by the adopted UPS reserve requirement. Benefits are strongly location dependent and influenced by workload flexibility, deferral duration, PUE, UPS energy and reserve requirements, and feeder loading. Under a combined adverse operating condition, FULL capacity decreases by only 3.4% relative to the nominal Mixed case while retaining substantial additional capacity over BASE.

In this study, we propose a novel rule-based interpretable choice model, Logic-Logit, designed to effectively learn and explain human choices. Choice models have been widely applied across various domains—such as commercial demand forecasting, recommendation systems, and consumer behavior analysis—typically categorized as parametric, nonparametric, or deep network-based. While recent innovations have favored neural network approaches for their computational power, these flexible models often involve large parameter sets and lack interpretability, limiting their effectiveness in contexts where transparency is essential. Previous empirical evidence shows that individuals usually use heuristic decision rules to form their consideration sets, from which they then choose. These rules are often represented as disjunctions of conjunctions (i.e., OR-of-ANDs). These rules-driven, consider-then-choose decision processes enable people to quickly screen numerous alternatives while reducing cognitive and search costs. Motivated by this insight, our approach leverages logic rules to elucidate human choices, providing a fresh perspective on preference modeling. We introduce a unique combination of column generation techniques and the Frank-Wolfe algorithm to facilitate efficient rule extraction for preference modeling—a process recognized as NP-hard. Our empirical evaluation, conducted on both synthetic datasets and real-world data from commercial and healthcare domains, demonstrates that Logic-Logit significantly outperforms baseline models in terms of interpretability and accuracy.

Geographic information systems increasingly rely on sophisticated spatial representation learning techniques to extract meaningful patterns from complex geospatial data. This paper introduces LEON, a novel self-supervised framework that adapts Graph Masked Autoencoders (GraphMAE) for geospatial region representation learning. Our method leverages the inherent spatial structure of geographic data by constructing hexagonal grid graphs using H3 indexing and applying masked autoencoding techniques to learn robust spatial embeddings from OpenStreetMap (OSM) amenity distribution patterns. We evaluate LEON on multiple real-world datasets including EuroSAT satellite imagery classification and various geographic prediction tasks (housing prices, crime prediction, and urban analytics). Experimental results demonstrate that LEON achieves significant improvements in spatial understanding, with up to 1.87% accuracy improvement on EuroSAT classification and consistent performance gains across geographic prediction benchmarks. The learned embeddings exhibit highly structured and distinct properties, making them particularly suitable for downstream spatial analysis tasks. Our findings suggest that self-supervised learning provides an effective paradigm for geospatial region representation learning using widely available crowdsourced data.

Vision-language-action (VLA) models have shown strong potential as generalist robot policies, but adapting them to unseen tasks often requires costly parameter updates. Recent work such as RICL introduces in-context adaptability by retrieving expert demonstrations based on the current VLA observation and providing them as additional context at test time. The effectiveness of this adaptation therefore depends critically on the retrieval mechanism. In this work, we systematically study how different retrieval methods affect both retrieval quality and task performance within the RICL framework. Specifically, we compare four different methods: image-based retrieval, retrieval augmented with VLA's state, retrieval using features from the VLA backbone, and random retrieval. Our experiments yield three main findings. First, no retrieval method consistently dominates the others in task success, while surprisingly, random retrieval achieves a non-trivial success rate. Second, standard retrieval-quality diagnostics do not reliably reflect downstream VLA performance. Third, demonstrations from different but related tasks can provide useful transferable information. Together, these results provide an initial step toward understanding how retrieval mechanisms shape the in-context learning capability of VLA models and their downstream task performance, while highlighting the need for more careful design and evaluation of retrieval mechanisms for reliable test-time adaptation.

Efficient and robust tracking of surgical robotic instruments is important for robot-assisted minimally invasive surgery, yet remains challenging due to the complexity of surgical scenes and the unconventional geometry of surgical instruments. Keypoint-based approaches are efficient, but their performance depends on reliable feature detection. Improving these detectors with real-world supervision is difficult because accurate real-world annotations are costly to obtain at scale. To address this limitation, we introduce a tracker-guided self-training framework that adapts a model pretrained on synthetic images to unlabeled real-world videos. Given measured robot joint states, an uncertainty-aware EKF recursively corrects the instrument pose and the observable joint angles by comparing projected model features with detected keypoints, shaft boundaries, and mask-derived cues. An RTS smoother subsequently refines the resulting trajectory, which is projected into pseudo-labels for fine-tuning the feature detector without laborious pose annotations. Experiments on real-world videos demonstrate consistent improvements from self-training across all evaluated keypoint metrics, and the resulting model outperforms prior approaches in both accuracy and runtime. The code and data will be released upon publication.

Generative flows sample from an unnormalized target by training a flow to be balanced, and the training loss is the signal a practitioner watches. We ask what that signal is worth: whether a small loss certifies an accurate sampler, whether the loss can be driven to zero, and how fast gradient descent does so. The loss decides the first. Losses that compare the two sides of the balance by their difference bound, in total variation, the error of the sampler the flow implies, with explicit constants that do not involve the policy; flow-matching losses that compare them through a ratio admit no such bound, already on a single cycle, whenever their generator is continuous at balance. On graphs, the backward policy decides the other two. Once it is frozen, balance becomes invariance under the backward chain, so that existence is free on finite graphs, and one constant — the norm of that chain's Green operator, which plays the role of an inverse spectral gap — fixes the order of the curvature of the loss around the balanced flow, from above and below, and sets a floor under the rate at which training converges near it. The mechanism is that gradient descent diffuses the flow along the backward policy. For the squared-logarithm generator of detailed and trajectory balance, training the balance loss on states converges globally on every finite path-connected graph, from every positive initialization. The constant can be infinite while backward trajectories are short on average, and exact flow matching can then fail. The bounds and rates are tested by exact computation on enumerable state spaces, and every theorem carries a certification status computed from a Lean~4 development.

Skill Retrieval (SR) aims to identify the most relevant skills from external skill libraries, and becomes increasingly challenging as libraries grow in scale and diversity. Existing methods either rank skills independently or rely on predefined graph propagation and hierarchical routing, making them vulnerable to semantic distractors, local trapping, and early routing errors. We formulate SR as an adaptive information-foraging process that coordinates region-level navigation with skill-level selection according to the utility and uncertainty observed during search. Based on this formulation, we propose SkillGATE, a graph-guided hierarchical retrieval framework with Gate-Aware Monte Carlo Tree Search (MCTS). SkillGATE constructs a graph-preserving hierarchical index and performs adaptive retrieval through selection, expansion, simulation, and backpropagation. G-PUCT guides action selection, expansion explores new regions, simulation evaluates candidate skills, and backpropagation updates search statistics. Experiments on six SR benchmarks show that SkillGATE consistently improves diverse retrieval and reranking backbones, achieving a 16.3% improvement in overall R@1 over the strongest retriever-based baseline. Our code is available at https://github.com/Edwinbe/SkillGATE-v1/.

Softmax attention gives every token a nonzero weight, which in trained models concentrates into attention sinks and massive activations that widen the dynamic range low-precision inference must cover. Softpick removes this constraint by rectifying scores, eliminating sinks and lowering hidden-state kurtosis, but its advantage fades at scale. We reframe this failure as a normalization problem. Softpick's denominator splits into positive- and negative-shifted sums $D^+$ and $D^-$, used identically in the forward and backward pass, preventing their roles from being isolated. We separate them into a family of operators that independently choose each denominator. The failure originates at initialization: every layer contains rows where $D^+$ is exactly zero, while near-dead rows produce gradient norms above $10^{12}$ regardless of the backward denominator. Only Softpick and a stop-gradient variant, which keeps $D^+ + D^-$ forward but backpropagates through $D^+$ alone, train from scratch. At 230M parameters, the stop-gradient operator matches Softpick on quantization, has fewer dead heads, and retrieves passkeys more reliably, trailing only on peak attention-weight kurtosis.