Papers

Sliding-scale insulin is the default hospital response to high blood glucose. Its correction factor was designed for patients who make no insulin, yet it is routinely ordered for patients whose pancreas still works, including those on high-dose steroids. We treat the sliding scale as what it is, a proportional controller sampled every four hours, acting on a patient who already contains a working glucose controller of her own. We test it in a non-diabetic adolescent on high-dose corticosteroids and parenteral nutrition, and in 4,995 virtual patients. It fails for three structural reasons, and no choice of correction factor fixes any of them. First, the patient's own loop settles in 11.5 minutes while injected insulin takes 64 minutes to peak, so most of each dose simply replaces insulin she would have secreted; in simulation the ordered scale moved her mean glucose by less than 1.5 mg/dL. Second, one unit lowers glucose twice as much at 110 mg/dL as at 190, so any correction factor strong enough to work when glucose is high overshoots when it is low; in 99.5% of virtual patients no constant factor does both. Third, four-hour sampling is far below the Nyquist rate, so successive readings are uncorrelated and each dose responds to a value that does not persist. Even the argument that correction doses rest the beta cells, taken on its own terms, favors continuous delivery. We propose replacing the scale with a continuous insulin infusion co-delivered with the dextrose and adjusted once a day on mean glucose, the one quantity four-hour monitoring can measure.

Egocentric human data offer a path to scaling robot learning beyond costly robot demonstrations, yet the embodiment gap makes raw human trajectories a poor supervisory target for control. Our key insight is that, although low-level actions are embodiment-specific, their underlying motion intent can capture task-relevant structure that transfers across humans and robots. We introduce EgoLAP, a VLA pre-training framework that jointly learns from human and robot trajectories through a shared language-based action chain-of-thought. EgoLAP expresses motion intent as structured, temporally abstracted language actions and pairs them with motion-level reasoning grounded in scene geometry, physics, and object affordances. Across extensive real-world and simulated experiments, EgoLAP transfers human experience to robot control more effectively than alternative action representations and reaches 80.1% mean real-world task progress, a 2.3x performance gain over alternative action representations. Motion-level reasoning also outperforms a composite reasoning format that combines subtask, object-box, and visual-trace reasoning.

Many long-horizon agents compact their context on a global rule, usually a token budget, blind to what the agent was doing. We ask whether the agent's recent behaviour predicts when a compaction will hurt. TRACE's public corpus of 590 harness-triggered AppWorld compaction boundaries replays each boundary from a re-executed prefix state under the pre-compaction context and under the summary, and records the burden of the next actions: calls that error or repeat a call already made. We find that pre-boundary history predicts post-compaction harm only weakly. An internally prespecified contrast by prefix placement is a wide null, and the naive "has-written" label behind it turns out to measure trajectory phase. The best extension-protocol trigger reaches held-out AUROC 0.66 (0.64 on the replicate's own label) against a same-boundary replicate of 0.72; the best frozen, interpretable trigger avoids 21% of harmful (positive-burden) boundaries while keeping 84% of compaction opportunities, and exceeds the random-rule expectation on count but not on burden mass (a post hoc comparison). Whether the best trigger beats a token-budget rule at matched retention cannot be evaluated on the release. We state what corpora should ship to answer it.

LLM-based agents are increasingly advancing scientific and engineering problem solving, with physics simulation emerging as a challenging yet practical testbed for reproducing complex physical phenomena with application in embodied AI, games and films. As the workhorse of such simulation, a solver computes how the state of a dynamic system evolves over time. Building such solvers requires physical understanding to identify appropriate models, mathematical reasoning to formulate the underlying dynamics, and software engineering to implement them as executable code, yet this capability of LLM agents remains underexplored. To this end, we introduce WorldSolver, a benchmark of 168 simulation tasks derived from physical phenomena in 61 classic computer graphics papers, spanning 7 physical domains. Each task contains a code scaffold that provides a fixed simulation environment for the scene, with the solver implementation left for the agent to complete. Specifically, we evaluate them along three dimensions: Execution Checks for successful execution, Visual Fidelity for reproducing the intended dynamic behavior in the rendered simulation, and Physical Plausibility for physics-grounded verification of the generated dynamics. Experiments on frontier agents reveal that producing executable solvers is difficult itself, and satisfying visual and physical correctness is even harder. GPT-5.6-Sol and Claude-Opus-5 perform comparatively better than the other evaluated agents, yet achieve overall scores of only 48.7% and 46.7%, respectively. WorldSolver is an early step toward agentic solver generation, and we hope it helps drive progress toward agents that can faithfully simulate the dynamic physical world. Code is available at https://github.com/sirujiang/WorldSolver.

Reusing the scores that select a Best-of-$N$ winner can overstate its expected reward. We study evaluation from a fixed matrix of $K$ independent scores per candidate for a policy that selects using $J$ fresh scores. A single estimator based only on this matrix is exactly unbiased for expected judge reward under every independent, stable collection of candidate-specific score laws if and only if $J<K$, for every pool size $M\ge N\ge2$. At $J=K-1$, the selector deepens as $K$ grows. For independent Gaussian scores with common variance and fixed $M\ge N\ge2$, the unbiased minimax risk in this regime is of order $\sigma^2/\sqrt K$, attained by Holdout; allowing bias improves the rate to $\sigma^2/K$. For two candidates, we derive the minimum-variance unbiased estimator at known variance and the sharp asymptotic unbiased minimax constant $1/(\pi\sqrt2)$, which Holdout attains without knowing the variance. The cyclic average over subsets and ties can be computed in $O(MK\log M)$ operations. At fixed selector depth, cyclic evaluation of bounded scores has $O(K^{-1})$ risk uniformly in pool size. The impossibility result concerns the fixed matrix: one additional fresh winner score permits unbiased evaluation of the all-$K$ policy.

Knowledge that a language model appears to forget during finetuning often remains stored and can be recovered, a phenomenon called spurious forgetting. Finetuning on new facts can even produce forgetting that undoes itself: recall of the old facts collapses, recovers as training continues on new facts alone, and only then erodes for good. We seek to understand when such forgetting is not catastrophic. A minimal associative memory reproduces these dynamics with three ingredients: keys with shared structure, concentrated new values, and normalization in the network. Finetuning moves all old representations along a common direction, hiding the old facts while preserving their relative geometry; normalization withdraws this shift once the new facts are learned, whereas fact-specific changes accumulate and cause the erosion. Moreover, subtracting the common shift eliminates the collapse in a Transformer trained on synthetic data, and removing a single direction from each weight update restores old facts in a pretrained language model. Forgetting thus combines a shared, reversible loss of access with a slow erosion of individual facts, and only the second is catastrophic. Which one dominates depends on whether the new data move old memories together or apart.

We study path-flow alignment as a unified training objective for flow matching. Instead of fixing the interpolation path and learning only the velocity field, we jointly train an endpoint-preserving path network and a flow network using the same alignment loss: the flow learns to match the path velocity, and the path learns to align its velocity to the current flow. Although every fixed learned path defines a valid flow-matching objective, the alignment loss alone is not a reliable criterion for path learning. We identify path overfitting, a failure mode in which the alignment loss decreases while sample quality worsens. We find that this failure is associated with low-entropy bottlenecks in the induced probability path, where the learned path routes samples through overly concentrated intermediate marginals. Motivated by this diagnosis, we introduce a stochastic path regularizer that hides part of the source information from the path network while preserving exact endpoints. The resulting regularization gives an explicit entropy floor for the stochastic training-path marginals and empirically suppresses the bottleneck in the learned sampler, making joint path-flow training effective. On ImageNet-256x256 with SiT backbones, our method consistently improves FID across model scales, extends to model-guidance training, and leaves the inference-time architecture and sampler unchanged. Code is available at https://github.com/lizeyu090312/traj_opt_paper

Generative retrieval trains a language model to generate the identifier of a relevant document. Recent work replaces the autoregressive decoder with diffusion, but changes identifiers, training recipe and decoding at once, so differences cannot be credited to the paradigm. On NQ320K and MS300K, we train autoregressive, masked-diffusion and block-diffusion models with residual-quantised, product-quantised and random identifiers. With identifier length and training budget fixed, we decode each model in several ways. Decoding alone moves a diffusion model's Hit@1 by 6.6 to 13.7 points. Our reference diffusion decoding, generate-and-match, generates an identifier, then retrieves the closest corpus identifiers. The generated identifier is right for 14-21% of NQ320K queries. We test one-pass scoring to decode diffusion retrievers: the model reads a fully masked identifier once, and each document is scored by its codes' probabilities. It matches or beats generate-and-match in 11 of 12 settings. Autoregressive models still lead in Hit@1; on NQ320K, the lead comes from the model, not beam search. Starting from one sampled identifier, one-pass scoring removes 46-83% of masked diffusion's deficit to beam search; from generate-and-match, at most a quarter. On NQ320K, every paradigm largely memorises which identifier answers which query: random identifiers keep 83-90% of the Hit@1 of residual-quantised ones. There, product-quantised identifiers lead residual-quantised ones by 3.4 points in the autoregressive model and by -0.7 to +3.6 in diffusion models; across decodings, AR's gap exceeds diffusion's by 1.5-2.3 points, around our 2-point threshold. Paradigm comparisons must report each paradigm at its own recipe and best decoding.

The following motif is common in spatiotemporal settings: we have a sequence of covariate and label pairs observed for a relatively short, recent time period. We have access to unlabeled covariates over a longer time period. Data is observed over many spatial locations. For instance, crop yield might be observed over a large geographical area for recent years, but weather data (which is informative about crop yield) is available for a much longer period. The goal is to estimate, at each spatial location, the expected label (e.g., crop yield) in the future and provide a valid confidence interval for this value. The observed time period alone is too short for reliable estimates. Imputing missing labels with machine learning can cause substantial bias. Prediction-powered inference (PPI) can correct for this bias, but it relies on an i.i.d. assumption that breaks under our expected temporal dependencies. Heteroskedasticity and autocorrelation consistent (HAC) procedures account for temporal correlation, but have not been adapted to cases where some labels are imputed. We provide reliable point estimates and confidence intervals given: short labeled time series (across spatial locations), a longer unlabeled time series, and an imperfect predictor of labels given covariates. We show our method outperforms natural alternatives.

Embodied agents must reason about 3D space while the video is still arriving, answering questions as soon as they have observed enough of the scene. VLMs that incorporate 3D geometric priors achieve strong spatial reasoning, but they operate offline, i.e., the full video must be available before they produce an answer. Streaming VLMs process frames causally and decide for themselves when to respond, yet they lack explicit 3D representations. We present SpaTime, a streaming VLM that fuses causal geometry tokens into the language model at every frame, using only the frames observed so far. To supervise when the model answers, we propose a response-time loss that maps per-frame response probabilities to a differentiable expected response time and penalizes the distance from the ground-truth frame. For evaluation, we construct StreamVSTI-Bench and StreamVSI-Bench, streaming adaptations of VSTI-Bench and VSI-Bench. On StreamVSTI-Bench, SpaTime reaches 49.2% overall accuracy and reduces the mean response-time error by 66% relative to the strongest streaming baseline.

We present an exposition of an argument, discovered by GPT-6 Pro, that gives an optimal bound for the polynomial Littlewood-Offord problem. Namely, let $F$ be a degree-$d$ multilinear polynomial that contains $r$ degree-$d$ monomials involving disjoint sets of variables. Then, for i.i.d. Rademacher random variables $\xi_1, \ldots, \xi_n$, we have $\mathbb{P}[F(\xi_1, \ldots, \xi_n) = 0] = O_d(r^{-1/2})$. This improves upon the previous bound of $(\log r)^{O_d(1)} r^{-1/2}$ due to Meka, O. Nguyen, and Vu, and resolves a conjecture attributed to H. Nguyen and Vu. The key part of the proof is an estimate for the total influence of bounded-degree rational functions, which resolves a recent conjecture of Kothari, Kovacs-Deak, Wang, and Yang.

Image stylization with latent-diffusion models entangles two independently refined axes: what a region depicts and how it is depicted. Such pipelines expose only global controls, yet professional retouching demands deliberate, region-specific control. We lift two conditioning weights already present in a ControlNet + IP-Adapter stylization pipeline from global scalars to per-location spatial maps, yielding local, per-axis control of content and style in a single generative pass. Because the two weights act on disjoint pathways, adjusting them independently spans a 2x2 retouching vocabulary, from free regeneration to identity preservation. We validate that edits stay confined to the retouched region and that each weight predominantly steers its own axis. Our approach requires no retraining and drops unchanged into any such pipeline.

In K-12 mathematics tutoring, student-tutor dialogue provides rich evidence of learners' problem-solving processes and sources of difficulty. Learning analytics research increasingly relies on large language models (LLMs) to extract such information from dialogue for a variety of downstream tasks, including knowledge tracing, behavioral modeling, and diagnosis of student reasoning errors. However, the validity of these model-generated interpretations remains insufficiently understood. In this exploratory study, we examine the validity of LLM classifications of five student failure modes in mathematics tutoring dialogue using an operational diagnostic codebook: uncertainty, misattribution, operator selection, conceptual gap, and procedural slip. Across models, human-LLM agreement was moderate (kappa = .524-.597), while cross-model agreement was substantially higher (kappa = .755-.781; alpha = .769). These findings show that cross-model agreement can create a misleading appearance of correctness, challenging the assumption that consensus among LLMs constitutes evidence of valid learner interpretation. For learning analytics, the implication is clear: scalable labeling is useful only if the inferred constructs are valid, and model consensus cannot substitute for independent evidence of that validity.

Battery energy storage systems must respect strict state-of-charge (SOC) limits to prevent overcharge and deep discharge, a task complicated by measurement noise and costly-to-model nonlinear dynamics. Data-enabled predictive control (DeePC) resolves the costly modeling issue by predicting future behavior directly from data. However, its regularization robustifies only the prediction and not the constraints. Scenario-based DeePC (Scenario-DeePC) extends DeePC with the scenario approach, building constraint robustness fully data-driven from observed prediction errors rather than an assumed disturbance distribution. This paper presents the first real-world deployment of Scenario-DeePC, on a grid-connected battery system at the NEST research facility, whose SOC estimate exhibits abrupt, irregular recalibration jumps in addition to ordinary noise. Compared to standard DeePC, Scenario-DeePC achieves comparable tracking performance with substantially fewer constraint violations. Its adaptive scenario buffer further tightens constraint handling automatically, improving robustness to unpredictable recalibration events.

We prove that the probability that Gaussian elimination with partial pivoting on $n \times n$ random matrices has growth $\rho$ is at least inverse quasi-polynomial: $\Omega(\exp(-c \log^2 (\rho)\log(n)))$ for some constant $c > 0$. This lower bound breaks standard conjectures in the smoothed and average-case analysis of Gaussian elimination. To the best of our knowledge, it is the first non-trivial lower bound on the probability of large growth for Gaussian random matrices.

Extended reality (XR) is increasingly a setting for empirical visualization research, where study-relevant state is distributed across headset and controller pose, scene configuration, selections, spatial layouts, and AR anchors. Existing XR tools support parts of the workflow, such as scene authoring, interaction, or session analysis, yet seldom treat an XR stimulus as a reusable component of a complete study lifecycle. We present reVISit-XR, an extension of reVISit that makes customizable WebXR stimuli embeddable, trackable, and replayable within empirical visualization studies. reVISit-XR sequences XR scenes alongside standard study components, collects reactive task responses, captures scene-authored semantic state together with generic XR traces, and rehydrates participant sessions for later desktop and headset analysis. Organized as a reusable stimulus build package and a study integration package, it lets XR stimuli act as first-class study components. We demonstrate its scope and feasibility through seven integrated, reusable examples and a deployed study, and discuss its current capabilities and future extensions.

Assistants from 2011 answered and waited, and agents from 2023 did a task and stopped. In September 2026 Meta's Muse showed an agent for one person, with accounts, devices, memory and a conversation that lasts, closed, in a vendor's cloud, in one country. Such an agent is expected to act on a person's accounts and devices, remember them across weeks, speak first when it is worth it, and answer for what it did. It is a kind of software, not a model, and until now had no open counterpart. This report defines the personal agent in five questions and three horizons. It reads how Muse is built from Meta's public record and a copy of its production prompt, each statement marked by its source. It then presents nanoMuse, the open-source counterpart under the GPL-3.0, one agent on every device a person owns, with hands on the phone's screen and the computer's. They share one conversation over a relay anyone can run; every action goes through a Sentinel, memory is files the person can read, and the model is their choice. Its size and cost are given as estimates. What is open, memory with provenance, an evaluation suite for the hands and an open model for them, is set out as a roadmap.

Large-model-based point-to-CAD generation holds immense potential for advancing industrial design and enhancing 3D modeling efficiency. However, most existing methods approach the problem as a general point-cloud encoding and token prediction task, neglecting the tokenization and supervision specifically for CAD-related primitives. As a result, these methods often struggle to accurately reconstruct the intricate primitive structures. To address this limitation, we propose PrimitiveCAD, a novel multi-stage paradigm for point-to-CAD reconstruction that enhances the geometric accuracy of generated CAD models while better preserving critical geometric features. First, we introduce a primitive-aware point cloud tokenization model, enabling the system to learn more robust geometric representations from CAD point clouds. Next, we perform supervised finetuning on a large language model (LLM) and introduce an operation alignment loss to align key CAD operation frequencies, thereby improving the preservation of global shape features. Finally, we incorporate reinforcement learning (RL) and introduce a feature-line alignment reward to further reduce stochasticity and enhance the fine-grained preservation of geometric features. Experiments on the DeepCAD and Fusion360 datasets show that our method achieves state-of-the-art performance in code validity, geometric accuracy, and geometric feature preservation.

Flying a quadrotor through a cluttered environment requires not only planning a collision-free reference trajectory based on perceived obstacles, but the reference also needs to be dynamically feasible and within the actuation limits of the vehicle, so that the controller can track it precisely. Existing methods either optimize a smooth polynomial inside a convex corridor, which limits agility, or treat obstacles as soft costs traded against tracking performance. We propose a Nonlinear Model Predictive Planning (NMPP) that imposes perceived obstacles as hard geometric constraints and hands a full-state reference to an obstacle-blind SE(3) controller. Our planner achieves a 58-67 % lower position RMSE than a linear Model Predictive Control trajectory planner and a 41-70 % lower RMSE than a polynomial trajectory planner. It also completes all forest flights with up to 9.5 m/s speed without collisions, and achieves 86 % flight success rate under a more aggressive speed profile where a state-of-the-art planner has only 26 % success rate. The real-world deployment showed reliable execution flying up to 5.5 m/s in an unknown cluttered environment.

Gene expression is widely measured in biomedicine, yet clinical outcome prediction remains challenging due to high dimensionality, strong feature correlations, and limited labeled data. Large self-supervised transcriptomic foundation models often fail to outperform simple supervised baselines. Tabular foundation models offer an alternative through in-context learning, but are typically pretrained on generic synthetic data rather than transcriptomic structure. We ask whether transcriptomics-aware pretraining, rather than scale, is the missing ingredient. Towards this end, we introduce GeneICL, a 4.2M-parameter tabular foundation model combining a semi-synthetic pretraining prior built from measured bulk expression profiles with a parameter-efficient recurrent architecture. We further enable right-censored survival prediction via a training-free reduction to regression using Cox partial-likelihood residuals. We evaluate GeneICL on 80 clinical outcome-prediction tasks spanning classification, regression, and survival. Tabular foundation models consistently outperform self-supervised transcriptomic models, while GeneICL achieves the best overall rank among evaluated foundation models and tuned baselines. GeneICL does so with up to 387$\times$ fewer parameters, no gradient updates at inference, and predictions within seconds on a laptop CPU.

Large language model agents are accelerating scientific automation, yet verified executions rarely become persistent program-level improvements, and existing evaluations do not examine this process across sequential tasks in both the natural and social sciences. We formalize ScienceClaw as fixed-parameter program self-evolution that unifies task solving, scientific verification, and program updates. ScienceClaw-Eval spans 23 disciplines and measures scientific correctness, evolutionary gain, retention, cross-dataset transfer, and evolution cost through sequential streams and independent reset evaluation. Our framework repairs executable workflows through multi-turn interaction, converts re-execution-verified failure–success trajectories into linked Skill and Operator candidates, and retains an update only when source-task replay reproduces the repair and independent scientific tasks improve. Code is available at https://github.com/beita6969/ScienceClaw.

Counterfactual inference in Gaussian-process structural causal models (GP-SCMs) has been developed primarily for continuous endogenous variables, limiting applicability to causal graphs that contain discrete child nodes with continuous parents. We introduce a unified probabilistic framework for counterfactual inference with heterogeneous variable types by pairing GP predictors with explicit exogenous noise mechanisms. For discrete outcomes, we derive exact conditional noise-abduction procedures using a uniform threshold for binary variables, a Gumbel-max race for nominal categories, and a latent Gaussian cut-point model for ordinal ones. In each case, we propagate abducted noise through interventions while accounting for posterior uncertainty in the GP latent functions, and prove that the resulting mechanisms reproduce the fitted model's observational and interventional distributions. On synthetic SCMs with known ground-truth counterfactuals, we evaluate estimation accuracy, consistency, and robustness to coupling misspecification. A key finding is that applying a categorical coupling to ordinal data inflates counterfactual error roughly threefold even when observational fit remains comparable, and that this error does not diminish with more data. As the training set grows, the fitted structural equation converges to the truth while the counterfactual error flattens onto a floor. In the reverse direction, forcing a false order onto nominal data instead degrades the fitted equation itself. The choice of coupling must therefore be justified on structural grounds rather than read off the fit.

This work presents a framework for accelerating transformer-based language models (LMs) on resource-constrained IoT devices. The framework targets compact LMs: BERT-Tiny (B-Ty), MobileBERT (M-Bt), MiniLM (M-Lm), Electra (E-Lt) and DeBERTa (D-Bt) – selected for their architectural diversity and use in edge inference scenarios. The proposed flow derives lightweight instruction set extensions tailored to the non-obvious computational patterns of these models. In addition, a custom instruction is introduced to accelerate the address generation stage of batch matrix multiplication, achieving a performance improvement of 15.39–21.74% with modest ASIC overheads of 6.79% in area and 2.33% in power. To further enhance performance without incurring additional processor core hardware cost, an optional compiler-directed loop unrolling strategy is employed, trading increased code size for overall reduced execution time. Evaluation on the Synopsys trv32p3f RISC-V core demonstrates inference speedups of up to 2.19x, and FPGA implementation on the AMD Zynq UltraScale+ ZCU102 shows a 32.94% area overhead at 75 MHz, whereas the ASIC implementation using the TSMC 28 nm library incurs a 21.08% area overhead while operating at 250 MHz.

Modern video models can generate realistic videos from real appearance references and proxy renders that specify scene structure, viewpoint changes, and motion. Evaluating this render-to-real capability requires a real target video depicting the same scene evolution, paired with an editable, geometrically registered 3D replica. Such data has traditionally required substantial manual modeling, calibration, and animation effort. We introduce RenderBench, a benchmark of 12 reconstructed real-world scenes spanning large-scale indoor environments and egocentric viewpoints, with both static and dynamic settings. Our construction pipeline combines visual geometry, neural reconstruction, and assisted 3D authoring. Each scene is decomposed into static objects and dynamic actors, registered to the capture cameras, and accepted only after multi-view geometric and temporal validation. Each evaluation unit contains appearance reference images, a held-out real target video, an editable digital twin, a matched proxy render, and renderer-native scene annotations. We evaluate transfer models against paired real target videos, retain PAI-Bench-C-compatible structural projections, and use scene annotations to localize failures by object, visibility, articulation, and motion. The first release retains 12 of 14 registered samples (85.7%), comprising 1,496 paired real-proxy frames. All released scenes pass file-integrity and environment-edit audits, while proxy diagnostics yield a depth si-RMSE of 0.2170 and instance mIoU of 0.3673. RenderBench provides paired real observations and editable scene state for assessing both appearance fidelity and preservation of geometry and dynamics.

Full-duplex speech models are trained to converse with a person, but they are increasingly made to converse with each other, in self-play data generation, agent societies, and model-based evaluation. In that loop no human absorbs a timing error: each model's turn-taking is the other's input. We ask what timing the loop settles into. Two PersonaPlex-7B instances exchange audio tokens on a shared clock in unscripted conversation, and one floor-transfer rule is applied to them and to Switchboard. Their timing is coupled: re-pairing speakers across conversations destroys it. But the floor changes hands late, at a median of 400-560 ms against 137 ms for humans, and the last 120 ms of the partner's turn, where human projection places a tenth of its transfers, holds 1% of theirs. Delaying one direction of the channel shifts the response one-for-one and leaves the run-up to it empty, consistent with a reactive wait after the perceived end rather than the turn-end projection human timing requires.

Visualization research has long examined embellishment and animation, while the design of rich interaction-contingent feedback remains under-articulated as a systematic design space. We introduce JuicyVIS, a theory-informed operationalization of juicy feedback for interactive visualization that translates ideas from game studies into a visualization-centered framework. JuicyVIS introduces three dimensions, interaction type, feedback timing, and feedback intensity, which we explore through 26 controlled prototypes, grounded in established visualization interaction categories and prior work on juicy feedback in games. We evaluate the juicy design space through three exploratory mixed-methods online studies. Results show that juicy feedback was consistently associated with higher engagement and aesthetic experience, while the clearest gains came from post-interaction feedback, not, for example, by simply increasing intensity. We discuss how juicy feedback can be thought of as a design material in visualizations, shaping how visualization interactions feel, how action consequences are perceived, and how users remain engaged with data representations.

Endpoint accuracy on abstract-reasoning benchmarks does not reveal whether a language model has acquired a transferable rule or fit distribution-specific regularities. We study this distinction in small language models on the ARC-TGI benchmark, which organizes abstract grid transformations into controllable task families and supports resampling, spatial shifts, and cross-benchmark transfer. Across more than 1,000 runs, we profile decoder-only, encoder–decoder, and mixture-of-experts model families under supervised fine-tuning. We examine the efficiency and stability of skill acquisition, robustness beyond the training distribution, interactions with model family and task formulation, and layer-wise attention signatures that accompany behavioral differences. Substantial in-distribution accuracy is attainable, but acquisition is sensitive to optimization and unevenly distributed across task families. Performance deteriorates sharply outside the training distribution, including when the rule is retained but grid scale changes. Greater training-set depth and breadth yield uneven gains, while the effect of additional in-context examples depends on model family. Executable-rule induction also yields correct solutions not observed under direct grid generation. On selected tasks, attention diagnostics show distinct concentration and context-dependence profiles, but do not establish general causal mechanisms. Overall, abstract-reasoning scores are conditional on the model, adaptation regime, evaluation distribution, and response format.

Speculative decoding accelerates inference for a large language model (LLM), referred to as the target model, by first using a smaller model, referred to as the draft model, to generate candidate tokens and then verifying them with the target model for acceptance or rejection. Prior studies primarily focused on the efficiency-utility trade-off of speculative decoding, e.g., lossy speculative decoding, leaving its security implications largely unexplored. In this work, we bridge this gap by providing the first systematic study of the security implications of speculative decoding. Through a large-scale measurement study, we reveal a pronounced security-utility asymmetry: across a wide range of lossy speculative decoding methods, improvements in inference efficiency come at a disproportionately high cost to security, with attack success rates for jailbreak and prompt injection attacks increasing much faster than utility degrades. We then propose SecureSD, a new theory-guided speculative decoding method that enhances security while maintaining efficiency and utility. Specifically, our theoretical analysis reveals that security degradation primarily originates from the early tokens generated by the draft model. Motivated by this insight, SecureSD applies a stricter verification criterion to draft-model tokens at early decoding positions. Extensive experiments on both security and utility benchmarks demonstrate that SecureSD significantly improves security while preserving efficiency and utility compared to existing speculative decoding methods.

Off-policy evaluation (OPE) for contextual bandit policies becomes challenging when action-level importance weighting incurs excessive variance. Doubly robust (DR) estimation remains unbiased under common support but retains these high-variance action-level weights. A prior estimator, Off-policy evaluation with Conjunct Effect Model (OffCEM), replaces them with more stable cluster-level weights, at the cost of relying on local correctness of the reward model. In this paper, we show that, under the assumptions required by DR and OffCEM, there exists an unbiased family of estimators that interpolates between OffCEM and DR. Building on this result, we propose the Variance Optimal-CEM (VOCEM) estimator, which selects the interpolation coefficient to minimize variance. We derive the population-optimal coefficient in closed form and show that the resulting estimator has variance no larger than either endpoint, OffCEM or DR. Experiments in controlled synthetic settings and on two large-action benchmarks show that VOCEM improves upon both endpoints in all 23 evaluated conditions, exhibiting greater stability and empirical robustness.

In this paper we consider matrices whose entries are rational functions of several parameters over a Moore-Penrose field, that is, a field with an involutory automorphism over which every matrix has Moore-Penrose inverse. We prove that over any such field the Penrose conditions can be solved with linear algebra alone, even though they form a polynomial system of degree two. We then bound the degrees of the numerator and of the denominator of the pseudoinverse in terms of the degree of the entries and of the rank, and not of the dimensions of the matrix. Specialization is the thread: the closed form is computed once over the field of rational functions, and the question is at which parameter values it still returns the pseudoinverse of the specialized matrix. We determine the values where it does not, from the matrix alone and before computing the pseudoinverse. They are the zeros of a polynomial built from the maximal minors, and they are the values at which the rank decreases. This turns previous sufficient conditions into an exact characterization. The theory also gives symbolic algorithms for the real, the complex and the parametric case, which we implement in Maple and test on 972 timed replications. We also apply them to a Leontief economic model, where the excluded value is the point at which the economy stops being viable.