Papers

Model explainability is essential for understanding and trusting machine learning models. Existing explainable AI methods often explain predictions through feature importance, counterfactual explanations, or rules. However, a region-level characterization of when and only when a prediction behavior arises remains less explored. This paper proposes Causal Sufficient and Necessary Regional Explanations (SNRE), a framework that learns an input region $A$ and output region $B$ such that membership in $A$ is both sufficient and necessary for the model output to fall in $B$. Motivated by the classical Probability of Necessity and Sufficiency (PNS), we formulate a region-level PNS measure through stochastic interventions and derive a differentiable finite-sample estimator for optimization. SNRE parameterizes the input-output region pair with explicit and interpretable algebraic region families, together with a learnable feature mask, balancing expressiveness and interpretability. Experiments demonstrate that SNRE learns region pairs with strong sufficiency-necessity performance, robust explanation behavior, and practical utility for model analysis.

We characterize when compressed vector representations admit exact linear or affine readouts of a finite lexicon's truth conditions: one fixed map per predicate, sending each entity vector to the corresponding truth vector. A necessary and sufficient row-space condition determines existence; the augmented truth matrix has rank r, giving minimum dimension r in the linear case, and r-1 in the affine. Exact readouts return values in a shared truth basis on which Boolean connectives act unchanged; separability alone requires an intervening threshold. For binary relations, exact bilinear readout of identity or strict total order requires linearly independent entity vectors. Experiments with GloVe and word2vec distinguish exact affine recovery, linear separability, and held-out prediction: most predicates are strictly separable, but none admits an exact affine readout from the pretrained embeddings. Supervised transductive training attains exact affine recovery to numerical precision at every tested dimension meeting the bound. At the embeddings' original dimension, geometries constrained to exact linear recovery retain 98-99 percent of the pretrained variance on the feature norms, and 80-83 percent on the WordNet lexicon.

We study the problem of scheduling jobs on a serial-batch machine with the aim of minimising the total weighted late work. In a serial-batch setting, jobs within a batch are processed sequentially, and none are removed from the machine until the last job in the batch completes its processing. The processing time of a batch is the sum of the processing times of the jobs within it, and the completion time for each job in the batch is equal to the makespan of the jobs in the batch. When a new batch begins, a constant setup time is required for the machine. We show that minimising the total weighted late work in this environment is $NP$-hard even if all jobs have a common due date and unit weight. For the general problem, we present a pseudo-polynomial time dynamic programming algorithm. Additionally, we explore two special cases, i.e., one with a common due date and another with an agreeable condition among due dates, processing times and weights. For both special cases, we develop specialised pseudo-polynomial time dynamic programming algorithms. The proposed approaches are equipped with specialised acceleration techniques to enhance their computational performance. The extended experiments demonstrate that the dynamic programming algorithms outperform Gurobi in time efficiency.

Recovering latent interaction structures from multi-agent dynamics is important for understanding and predicting interacting systems. Trajectory-based structural inference has achieved promising performance, but conventional formulations assume that the trajectories of all modeled agents are available. In practice, agents may become unobserved at deployment because of limited sensing, occlusion, or communication failure. Existing studies have considered unseen-node estimation, structural inference under partial observations, and missing-value imputation, yet the joint recovery of hidden-agent trajectories and their interactions remains underexplored. We formulate this problem as structural inference under hidden agents. Its key difficulty is a circular dependency: recovering interactions involving a hidden agent requires an estimate of its trajectory, while trajectory reconstruction can itself benefit from structural information. To address this challenge, we propose Structural Inference under Hidden Agents (SIHA), which combines structure-agnostic initialization with structure-guided iterative refinement. SIHA reconstructs hidden trajectories from visible observations, infers interactions using Neural Relational Inference, and feeds the estimated structure back into hidden-state reconstruction through multi-strength structural attention and iterative state–structure updates. Experiments on three benchmark dynamical systems demonstrate consistent improvements in visible-to-visible structural inference, while also showing benefits in hidden-state reconstruction and future prediction. Motion-capture experiments with simulated whole-limb occlusion further demonstrate its effectiveness in realistic hidden-agent settings.

Voice assistants grounded in external knowledge typically use automatic speech recognition (ASR) to transcribe speech queries before retrieving evidence from textual knowledge bases. This cascade adds latency and propagates recognition errors, whereas direct speech retrieval is vulnerable to cross-modal misalignment. To address these limitations, we propose DUPAR, a conversational retrieval framework with complementary slow and fast paths. The fast path uses a task-adapted audio encoder aligned with frozen BGE-M3 text embeddings to search a cross-turn evidence cache. When cache confidence is insufficient, the slow path fuses full-index retrieval using audio and ASR-transcript embeddings, and the selected evidence refreshes the next-turn evidence cache through one-hop graph expansion. On a domain-specific knowledge base, our trained audio encoder approaches text-retrieval accuracy on clean speech with a 3.75$\times$ query-side speedup over ASR + Text Encoder. It raises average Recall@10 from 0.771 to 0.875 on the noise benchmark and improves overall Recall@1 by 4.2 percentage points across synthesized speaking styles. Compared with full-index audio retrieval, cross-turn evidence caching significantly reduces retrieval errors when the previous turn retrieves correct evidence and the follow-up targets a one-hop neighboring chunk.

Multi-encoder fusion extends Large Audio-Language Models (LALMs) beyond speech-centric recognition, but selecting encoders via intuition or exhaustive search often introduces redundant representations and inflates an already constrained compute budget. We propose CUES (Correlation-gUided Encoder Selection), a lightweight heuristic that estimates complementarity through task- and category-level Pearson correlations between encoders' performance profiles, scoring a candidate set from single-encoder evaluations alone–without fusion training during selection. Evaluated on the XARES-LLM benchmark with a frozen SmolLM2-135M backbone (LoRA-adapted) via five-fold cross-validation, CUES consistently identifies the same configuration per track from held-out development splits alone, without using test data for selection. For the broad Track~A suite, CUES selects a cross-family trio (Whisper-medium, mHuBERT-147, and Dasheng-base), achieving a 4.3% relative gain over Whisper-medium (0.771 vs. 0.739). For Track~B text generation, it re-anchors on a focused, speech-only pair (mHuBERT-147 and WavLM-base-plus) and actively abstains from adding a divergent encoder, outperforming mHuBERT-147 by 6.3% (0.589 vs. 0.554). Rather than a failure to scale, this divergence is consistent with a diversity–interference trade-off that CUES navigates per track from correlation signals alone: across the evaluated pool, added cross-family diversity tends toward an inverted-U on broad audio tasks but toward steady degradation on text generation, which favors a focused, speech-anchored set.

The world population is growing rapidly, and technology is improving in parallel. Meeting the huge demand for food for these 7 billion people not only depends on increasing food production but also on reducing food loss. Crop losses due to disease affect both the food supply and the financial and economic stability of a country. Tomatoes are among the top food-producing crops globally, and a significant portion of this production is lost due to disease. People have used Machine Learning techniques for feature extraction and early diagnosis of tomato diseases, and nowadays, Deep Learning-based models are widely used for disease recognition. However, most existing models are highly parameter-intensive, which increases the time required for training and inference. As a result, while lightweight models are more suitable for user-friendly applications, they often show a reduction in performance. To balance performance and model size, we propose CoAtNet-DeepMoE, a Convolution-Attention hybrid architecture for rich feature extraction, further enhanced with a DeepSeek Mixture of Experts to substantially reduce the number of parameters without sacrificing accuracy. We evaluate our model on both balanced and imbalanced datasets from Kaggle and PlantVillage, demonstrating robustness and achieving 99.80% accuracy, 99.80% precision, 99.80% recall, and 99.80% F1-score on Kaggle, and 99.83% accuracy, 99.85% precision, 99.76% recall, and 99.80% F1-score on PlantVillage, representing state-of-the-art performance with only 2.47M parameters. The source code will be available at https://github.com/nadimbrur/CoAt-MoE.

Segment Anything Models provide reusable priors, yet they require user prompts and cannot support fully automatic instrument segmentation. Automatic prompting is difficult for thin, articulated, reflective, and partly occluded tools, where several configurations can be valid. We formulate automatic prompting as lightweight point-set planning and isolate the point source under a frozen pathway. To this end, we present SetPlanner, a 1.52M-parameter plug-in point-set planner for frozen SAM. The plug-in preserves SAM's point-prompt interface and enables reuse across backbones. SetPlanner plans complete unordered K-point sets from geometry-aware targets with a permutation-aware conditional flow. SAM decodes eight candidates; their consensus readout yields a ground-truth-free prediction. Across three endoscopic datasets, SetPlanner wins all six transfer routes over a LoRA-adapted system. Under our frozen-pathway protocol, SetPlanner reaches 0.934 Dice on Kvasir-Instrument and recovers 96% of a 44.4-point localization gap, while candidate disagreement ranks low-Dice cases at AUROC 0.969.

Novel view synthesis from unposed multi-view images remains challenging, as the model must jointly learn scene representations and camera parameters without pose supervision. Existing approaches largely fall into two extremes: implicit latent-space rendering is flexible and easy to optimize, but often yields weakly grounded camera estimation; explicit 3D representations provide stronger geometric grounding, but introduce heavier parameterization and more fragile optimization. In this paper, we present IRIS, a fully self-supervised framework that provides a practical middle ground between these two paradigms. Instead of decoding free latent tokens or reconstructing fully explicit 3D primitives, IRIS represents the scene as a latent neural field and renders novel views by querying this field under self-predicted cameras. Specifically, projected features from reference views are aggregated at sampled 3D points to form point-wise latent features, which are then composed along target rays for rendering. This design preserves the flexibility and optimization stability of implicit modeling, while introducing stronger geometric structure than unconstrained latent rendering. Extensive experiments show that IRIS achieves strong novel view synthesis quality with competitive pose accuracy under fully self-supervised learning. Our project page: https://leo-frank.github.io/IRIS

The brain needs an accurate internal model of the world to generate predictions and guide behavior. However, it must estimate the statistical structure of the environment from limited experience. This is particularly difficult for rare events, whose observed frequencies in a limited sample may substantially under- or overestimate their true frequencies. How the brain constructs an accurate internal model despite this sampling problem remains unclear. We address this problem using a Bayesian Confidence Propagation Neural Network (BCPNN) trained on event sequences from a Markov-chain random walk with controlled event frequencies. Treating the underlying Markov structure as the ground truth, we train the network on limited sample of event sequences and then allow it to generate autonomous replay based on the learned structure. We evaluate replay fidelity at the levels of both marginal event frequencies and conditional transition structure. We find that moderate neural noise, modeled as temporally correlated random fluctuations in unit activity during replay, is critical for faithful internal simulation. Without this variability, deterministic replay systematically under- or overrepresents rare events, whereas moderate noise restores both their marginal and conditional occurrence. Moderate noise also broadens the range of parameter values that produce accurate replay, making the model more robust to parameter variation. Together, these results support noise-assisted internal simulation as a potential mechanism for compensating for sampling errors arising from limited experience. Our model also provides a testable framework for investigating how altered neural variability may impair internal-model fidelity in disorders such as Parkinson's disease.

Contact-rich physical human–robot interaction (pHRI) imposes time-varying demands associated with physical interaction, motor regulation, and physiological response, motivating continuous assessment of interaction workload. This paper presents an online multimodal assessment framework that integrates interaction wrench, planar tool-center-point (TCP) kinematics, and skin conductance level (SCL) into four interpretable workload-related factors. Their relative contributions are adjusted using path curvature to reflect changes in motion demand and task progression to account for gradual physiological variation over time. The framework was evaluated with 24 participants across 18 controlled combinations of temperature, acoustic noise, and illuminance under two admittance-control modes. Strict leave-one-subject-out (LOSO) evaluation used standardized pupil diameter ($\mathrm{PD}_z$) as an independent physiological reference and included comparisons with static variants and representative state-of-the-art learning-based baselines. The proposed framework achieves a cohort-mean $30\,\mathrm{s}$ block-wise Spearman correlation of $\rho_{30}=0.308$ with the physiological reference, with positive subject-level correspondence in 23 of 24 participants. Its overall performance is comparable to the state-of-the-art learning-based baseline. At the same time, our framework keeps the assessment process transparent through explicit workload-related factors and defined weighting rules, while outperforming the corresponding fixed-weight formulation. The framework also maintains consistent performance across the two tested admittance-control modes. These results support a transparent and interpretable approach to continuous interaction workload assessment in contact-rich pHRI.

AI is increasingly used to support decarbonization decisions across the built environment, yet the development, training, and use of AI consume energy and induce CO2e emissions. However, existing assessments often report physical-system savings while omitting AI-side emissions. Moreover, they rarely account for the mismatch between when AI costs occur and when decarbonization benefits materialize, which may be substantial for infrastructure-scale projects. To address these issues, we present a time-aware assessment framework that models avoided emissions and AI-induced emissions as discrete-time streams over a finite time horizon. In demonstrating this process, we seek to show that time-aware assessment can support temporal decision-making, identify cases in which accounting for time value of carbon can change preferred rankings relative to time-invariant totals, and explore how decisions may vary with slightly different governance priorities. Using four representative interventions with intentionally different temporal profiles (multi-project low-carbon concrete design support, AI-assisted construction logistics, agentic HVAC control, and predictive maintenance), we demonstrate how discounting can change preferred rankings relative to time-invariant totals and supports ranking sensitivity analysis, discounted payback screening, and break-even discount-rate analysis. We also provide decision guidelines that support go/no-go screening, timing decisions, and minimum "bang-for-your-buck" thresholds. Ultimately, this work contributes a lightweight framework for deciding whether and when to deploy AI-enabled interventions for decarbonization under explicit time preference.

An agentic framework is proposed for autonomous site-specific base station (BS) deployment in wireless network planning. In contrast to conventional approaches that rely on manual site surveys or extensive ray-tracing (RT) simulations with significant human intervention, the proposed framework autonomously explores and optimizes BS deployment under a limited RT evaluation budget, enabling resource-efficient network planning. To this end, a continuous, geometry-grounded deployment action space is first constructed from three-dimensional (3D) wireless digital twins. Within this action space, an agent team operates through a stateful perception–reasoning–reflection loop. Specifically, a Placement Agent first generates candidate BS deployments in two complementary modes: an experience-guided mode that refines promising solutions, and an exploration mode that avoids getting stuck in local optima. After the candidate deployments are evaluated through RT, a Reflection Agent interprets the RT results together with the scene geometry, identifies performance-limiting factors such as blockage, overlapping coverage, and uncovered areas, and converts these diagnoses into guidance for subsequent deployment. Through this iterative process, site-specific experience is accumulated and deployment plans are optimized without human intervention. Numerical results in two realistic urban scenarios show that: 1) the proposed approach substantially outperforms heuristic, learning-based, and large language model (LLM)-assisted methods; 2) it achieves highly competitive coverage against the optimal solution while requiring substantially fewer transmitter-level RT evaluations; 3) site-specific reflection effectively turns raw RT feedback into refinement guidance, whereas the dual-mode mechanism preserves diversity and facilitates escape from local optima.

MLegS is an open-source Modern Fortran code package for solving linear and nonlinear time-dependent partial differential equations on radially unbounded domains. It builds on an earlier in-house prototype, recasting it as a modular solver with a unified interface for fields, transforms, and spectral operators. Within this framework, the numerical formulation maps $0\leq r<\infty$ onto a finite interval, where associated Legendre functions provide the radial basis, while Fourier series represent the periodic azimuthal and axial directions. The mapped radial basis represents radial infinity without imposing an artificial boundary condition at finite radius while incorporating the regularity required at $r=0$ according to azimuthal order. The implementation distributes the corresponding field layouts using the Message-Passing Interface (MPI), with strong- and weak-scaling measurements extending to 512 cores across four compute nodes. Beyond parallel performance, three examples assess the solver across diffusion, reaction-diffusion, and vortex dynamics: closed-form diffusion solutions establish spectral spatial convergence and the design orders of both semi-implicit integrators; a Fisher-Kolmogorov-Petrovsky-Piskunov front matches the linearized solution and propagates at the classical speed within its logarithmic correction; and vortex-pair simulations based on toroidal-poloidal decomposition yield a measured Crow growth rate $2.7\%$ above the theoretical prediction and demonstrate the reconnection of the vortex pair.

Strong chromospheric absorption lines such as H$\alpha$ 6562.8 A and Ca II 8542.1 A provide vital diagnostics of plasma dynamics and thermal structure in the solar chromosphere. Multilayer spectral inversion (MLSI) offers a physically interpretable framework for modeling these lines using a finite number of radiative-transfer layers, but conventional MLSI relies on pixel-by-pixel nonlinear least-squares fitting, making it computationally expensive for large imaging spectroscopic data sets. Here, we introduce a physics-informed neural-network (PINN) framework to accelerate MLSI while preserving its analytic radiative-transfer formulation. The network predicts MLSI parameters directly from observed line profiles and passes them through a differentiable MLSI forward model to synthesize spectra. Training follows a two-stage approach: an initial stage optimized solely via spectral reconstruction loss, followed by fine-tuning that combines spectral consistency with parameter-space supervision from conventional MLSI results on a single reference image. This strategy eliminates the need for large precomputed training sets while maintaining physical interpretability. Applied to Fast Imaging Solar Spectrograph (FISS) observations from the Goode Solar Telescope (GST) targeting both quiet-Sun and active-region regions, MLSI-PINN parameter maps reproduce the primary spatial structures of direct inversions, achieving an arithmetic mean pixel-wise Pearson correlation coefficient of 0.933 across all evaluated parameters. The reconstructed spectra closely match both observed profiles and conventional MLSI fits. Post-training, MLSI-PINN processes a raster in approximately 5-15 seconds compared to 3-5 minutes for conventional MLSI, delivering an inference speedup of about 12-60 times without substantial loss in reconstruction quality, enabling efficient MLSI analysis on large chromospheric data sets.

We study data structure lower bounds for the Approximate Matrix Powering (AMP) problem. Given a substochastic, symmetric matrix $\mathbf{M}\in\mathbb{R}^{n\times n}$ and parameters $k$ and $\alpha$, the goal is to preprocess $\mathbf{M}$ so as to answer entry queries $(u,v)\mapsto \mathbf{M}^{k}[u,v]$ up to additive error $1/n^{\alpha}$. We focus on AMP in the succinct and systematic regime, in which the data structure stores $\mathbf{M}$ verbatim, uses an additional $r$ bits of redundancy, and must answer queries by probing only a small number of entries of $\mathbf{M}$. Our main conceptual contribution is a general framework for proving probe–redundancy trade-offs for systematic data structures. We introduce the query-with-sketch model and develop a min-entropy-based approach that lifts conditional min-entropy bounds in the absence of redundancy to probe lower bounds in the presence of redundancy. We then establish these min-entropy bounds using problem-specific analytic and algebraic tools, for the downstream applications to AMP and its variants. As a consequence, our results provide new unconditional evidence toward a conjecture of Patrascu and Roditty (2010) on the space required for constant-time set-disjointness queries.

Chatterjee and Sloman proved that a bounded measurable similarity function with sufficiently small average Gromov hyperbolicity admits a tree representation with small mean approximation error. Their argument uses a weighted version of Szemerédi's regularity lemma and does not yield explicit quantitative bounds. Here, we establish a tighter relation between average hyperbolicity and mean tree approximation error. For a similarity function $s:S\times S\to[0,b]$, we prove that $$\operatorname{Tree}(s) \leq (63/e)^{1/3} \sqrt[3]{b^2 \operatorname{Hyp}(s)} \leq 2.8512 \sqrt[3]{b^2 \operatorname{Hyp}(s)}.$$ The proof uses a simple pivoting construction inspired by \KwikCluster. We also discuss the optimal dependence on average hyperbolicity, including a square-root lower bound, and connections with ultrametric fitting.

RTL source-level debugging research requires benchmark artifacts that provide faulty designs together with precise change locations, executable test stimuli, and reproducible configurations. Available Verilog resources usually provide only a subset of these elements. We present VeriBugBench, a framework for constructing Verilog RTL debugging benchmarks through empirically grounded fault construction, LLM-based testbench enhancement, and execution-based retention. The mutation library maps recurring, multi-granularity repair patterns observed in RTL bug-fix histories to 19 executable inverse operators. For each project, an LLM generates a design-specific stimulus phase from the clean DUT and original testbench; the phase is composed with the original testbench for candidate execution. Applying the framework to 45 open-source projects yields VeriBugBench-v1.0, with 2,608 executable single-fault instances whose effects are observable at design outputs. Across the 45 projects, the assembled testbenches increase mean project-level fault observability from 36.01% to 39.54% and improve line coverage and execution-trace diversity on average. VeriBugBench provides versioned RTL variants, source-level ground truth, testbenches, and execution artifacts for evaluating RTL debugging methods.

A decomposable randomized encoding (DRE) for a function $f$ allows $n$ parties, using shared randomness, to encode their individual inputs locally so that the collection of encodings reveals $f(x_1,\ldots,x_n)$ and nothing else. DREs are widely used in efficient multiparty computation. Their main complexity measure is size, the total bit length of the local encodings. Yet the optimal DRE size remains poorly understood even for the $n$-bit OR function. We prove the first superlinear lower bound for OR and, more generally, for every non-periodic symmetric function. Under an additional symmetry assumption, we prove a sharp $\Omega(n\log n)$ lower bound for OR, matching the classic construction of Feige, Kilian, and Naor (STOC 1994). We also prove the first $\Omega(n^2)$ lower bound on DRE size for non-explicit Boolean functions.

World-action models (WAMs) have emerged as a promising paradigm for robot manipulation by jointly modeling future visual dynamics and robot actions. However, existing WAMs are trained predominantly on successful trajectories, making them prone to failure when real-world execution diverges from the learned dynamics. This issue is amplified in autoregressive WAMs, where execution errors become part of the causal history and continue to influence subsequent predictions. To this end, we introduce \method{}, a training-free framework that reformulates failure recovery as test-time scaling over causal histories. This formulation decomposes recovery into three coupled decisions: when to revise the causal history, where to recover a reliable history prefix, and which history configuration best supports subsequent execution. Specifically, \method{} realizes these decisions through three stages: 1) Progress-Aware Recovery Trigger detects persistent non-progress and triggers recovery only when the current execution state permits intervention; 2) History-Prefix Recovery identifies the unreliable history suffix, retrieves a historical anchor matching the current physical state, and reconstructs the causal KV state from the retained prefix while conditioning on the latest real observation; and 3) Hypothesis Verification compares the future continuations induced by complete-history, recovered-prefix, and full-reset hypotheses, and commits the best-supported hypothesis. Experiments in both simulated and real-world manipulation settings demonstrate consistent improvements in task success, while ablations confirm the contribution of each recovery stage.

This paper treats planar swarm transit along a fixed route when no ground station streams the reference in flight. One embedded leader stores the deployment point and the destination, generates the reference onboard, and drives its progress rate with a saturated coordinate of a Chua oscillator; the saturation keeps the rate inside a prescribed positive band, which gives explicit bounds on the reference arrival time. This leader broadcasts the one scalar rate to the other leaders; every leader adds it as a velocity feedforward and holds its station by single-pinned consensus on position error, while followers use only relative position and velocity feedback through a spring-damper network. For kinematic leaders and double-integrator followers on fixed graphs with ideal information exchange, the leader errors decay exponentially, the follower errors are bounded, and the whole swarm eventually stays inside the destination area whenever the final formation fits strictly inside it. Ten simulations with 30 agents compare the chaotic rate with a constant rate. The reference arrivals satisfy the analytical bounds, and the chaotic rate raises the error of a constant-velocity predictor at all three tested horizons, at the cost of a larger but bounded follower error. The analysis does not address internal link failures or model-aware observers.

We completely characterize the second-order coding rate region of the two-user Gaussian multiple-access channel at points in the relative interior of the sum-rate face, under maximal per-codeword power constraints. For any fixed average error probability $0 < \varepsilon < 1/2$, a nontrivial mixture of power splits reduces the sum-rate dispersion and strictly improves upon standard inner bounds based on a constant power split. Thus, a constant power split, although sufficient to attain the first-order capacity region, is insufficient for second-order optimality in this regime. Our achievability proof builds on the constant composition and coded time-sharing framework of Scarlett, Martinez, and Guillén i Fàbregas, with an optimization over admissible power splits. A matching second-order converse establishes the optimality of the resulting coded time-sharing strategy.

In collaborative tasks with asymmetric information, participants coordinate their understanding through interaction. We ask whether gaze provides evidence about grounding across two such tasks. Working from discrete behavioral annotations, we map HCRC MapTask (Anderson et al., 1991) and MUNDEX (Türk et al., 2023) into a shared partner/task/away vocabulary and compute gaze features around task-relevant dialogue units. In both corpora, aligned reference interpretations (MapTask) and UND (understood) judgments (MUNDEX) are associated with more task-directed gaze and with less partner-directed gaze, lower gaze entropy, and fewer gaze transitions. The associations are clearest for the participant leading the task: in giver-produced references, and in explainer judgments, which also co-vary with the explainee's gaze. In same-speaker MapTask reference chains, the speaker's gaze entropy is lower at the mention where a previously non-aligned referent becomes aligned. The best gaze feature groups improve modestly over controls under grouped cross-validation: temporal features in MapTask and raw proportions in MUNDEX. Because effects are small and several weaken when recurring participants rather than dialogues are the unit of inference, we treat gaze as one contributing cue to grounding, to be interpreted alongside task and dialogue context.

Lightweight audio-visual speech enhancement (AVSE) models face a critical trade-off between computational efficiency and cross-modal alignment accuracy. While simple concatenation lacks relational expressiveness, dense cross-attention incurs computational overhead and is prone to unreliable cross-modal correspondence under strong acoustic interference. We propose Sparse Graph-Guided Mamba (SG-Mamba), a lightweight AVSE framework that integrates a sparse heterogeneous graph with a linear-complexity Mamba backbone. The graph explicitly models modality-specific relations through content-adaptive attention and cross-frame audio-visual connections, while Mamba captures long-range temporal context. We further introduce an audio skip connection to preserve spectral detail without sacrificing noise suppression. Evaluated on LRS3, SG-Mamba achieves competitive or superior performance against strong lightweight baselines and reaches 13.091 dB SI-SDR under noise-only condition. It also remains robust in cluttered multi-speaker conditions with a competitive cost of 3.45 G MACs (or 6.90 G FLOPs). Results on VoxCeleb2 further suggest that explicit structural priors improve robustness, generalizability, and computational efficiency in lightweight AVSE.

Text-to-image generative models are widely used in professional and creative settings, yet how they represent gender across occupations – and whether newer models are fairer – remains poorly understood across multiple generations. We evaluate gender representation across 20 occupations, 5 prompt templates, and 4 Stable Diffusion model generations (SD 1.5, SD 2.1, SDXL, SD 3 Medium), generating 8,000 images with n = 100 per occupation-model cell (5 prompts x 20 images), and classifying all with DeepFace. Across the 8,000 open-source images, 76.4% show male subjects (95% CI [75.1%, 78.7%], p < 2.2 x 10^-16, Benjamini-Hochberg adjusted). More strikingly, 57.6% of images for historically female-coded occupations show male subjects (raw p = 3.43 x 10^-22, BH-adjusted p = 1.71 x 10^-21). All nine significant tests reported in this paper survive BH correction across 10 tests. When compared against U.S. Bureau of Labor Statistics workforce data, models underrepresent women by 20-46pp on average, with particularly large deviations for near gender-balanced occupations: scientist (48% female in BLS, 82-99% male in model outputs) and cleaner (46% female in BLS, 80-92% male in outputs). Model generations do not improve steadily: bias worsens from SD 1.5 to SDXL before partially recovering in SD 3 Medium. A preliminary comparison with GPT-image-1 on five occupations suggests lower bias than open-source models, though the practical effect is small (Cramer's V = 0.080) and the comparison is exploratory. No model achieves gender parity.

Large language model optimization is an active research area, spanning quantization of model weights, early-exit methods for skipping layers, and speculative decoding. Each track uses its own quality measures, typically an idiosyncratic benchmark score. Few approach the measurement precision required by other scientific disciplines. We propose a rigorous methodology for measuring output quality, suitable for cross-system and cross-technique comparison. We score outputs with an LLM as a judge, but calibrate the judge formally: we compare its scores on two ordinary runs of a model given the same prompts, verifying that it shows no systematic preference between statistically equivalent outputs and measuring its per-sample noise. Each design also includes a 'null' condition, provably identical in distribution to the unmodified model, whose measured difference must be zero. With this one instrument we measure several acceleration techniques on the same prompts, so their quality costs can be compared. Perceived quality proves highly dependent on the domain of discourse. A 4-bit model was indistinguishable from its 16-bit original down to our design's +/-0.3-point resolution, in English prose and Chinese alike. At 3-bit precision the same prompts lost 0.5 points in English prose, 0.9 in Chinese, and 1.1 on multi-step math; early exit that cost 0.7 points on prose cost 2.5 on math, cutting correctly solved problems from 19 of 27 to 6. The pattern held for models from Alibaba and from Meta, but not its magnitude: the same quantizer cost Meta's model 1.8 points where it cost Alibaba's 0.7. A model's certainty about a token predicts how likely it is to differ from the full model's choice, but not how much that difference affects judged quality, so acceptance rules relying on certainty cannot distinguish errors that matter from errors that don't.

Active imitation learning reduces expert effort by allowing a learner to request the demonstrations it needs. Existing methods typically select these requests for their expected information gain about the expert policy. In structured multi-task domains, however, the number of start-goal tasks may grow combinatorially despite their solutions sharing reusable behavior. This makes composable behaviors especially valuable, since a single demonstration may help solve many tasks at once. Prior methods do not explicitly account for this value when selecting which demonstration to request. We introduce Adaptive Agents via Latent Topologies (AALT), which requests demonstrations that maximize expected gains in start-goal connectivity. We further show that this objective is formally tied to information gain about task reachability. AALT organizes existing demonstrations into a topology of latent hub states connected by learned behaviors, identifies high-value bridge demonstrations that are likely to enable many tasks at once, and grounds each to an expert query. At inference, it plans through the resulting topology and conditions a diffusion policy on each successive hub transition. In a simulated UR5e robot ordered-retrieval domain with 72 tasks, AALT improved from 42/72 to 72/72 (100%) successful tasks consistently using only 3 demonstrations totaling 5 transitions beyond the initial dataset. After 20 demonstrations, the strongest baseline averaged 88.6% success using 98 transitions.

The emergence of movable antenna technology enables flexible antenna placement, allowing transceiver antennas to more effectively exploit spatial degrees of freedom. In this paper, we investigate a monostatic near-field sensing system, aiming to minimize the worst-case squared position error bound (SPEB) over the entire near-field region by jointly optimizing the transmit and receive antenna placements, along with the transmit power allocation. Toward this end, we first derive the closed-form expression of SPEB by calculating the Cramér-Rao bound (CRB) for estimating the target's angle and distance. It is shown that the derived SPEB is governed by the power-weighted moments of the transmit antenna locations and the moments of the receive antenna locations, motivating the development of a moment-based optimization algorithm. Functional analysis of the SPEB's moment structure proves that the optimal antenna distributions can be realized by a centro-symmetric structure, and shows that the worst-case target is located at the array broadside on the Rayleigh boundary. Moreover, by leveraging moment-based analysis and the Richter-Tchakaloff theorem, we derive closed-form antenna distributions with three points supported at the aperture center and two edges. Specifically, the optimal transmit antenna distribution can be exactly realized by activating only three transmit antennas while the optimal receive antenna distribution consists of three clusters. Numerical results show that the proposed closed-form design significantly outperforms conventional antenna placements and power allocation benchmark schemes.

Selective state space models (SSMs) have recently emerged as a compelling alternative to transformers, combining competitive performance with substantially improved inference efficiency. At each SSM layer, a sequence of hidden states are propagated by a recurrence, mixing information of different tokens. Despite using a different mechanism, this mixing plays a role analogous to attention in transformers. In fact, recent works have shown that the two architectures may be closer than they first appear, as this recurrence admits a formulation akin to linear attention. In transformers, attention is known to drive the tokens to cluster, i.e., to reach consensus, collapsing in the limit to a single direction. Thus, we ask: does the recurrence at the core of SSMs drive the tokens to consensus, as attention does in transformers? To answer this question, we take a dynamical systems perspective on SSMs, modeling the evolution of tokens across layers as an ordinary differential equation. By exploiting input-to-state stability arguments, we establish local exponential stability of the consensus equilibria and characterize their domain of attraction for time-varying weight matrices, a setting not addressed by previous results. We thereby show that the resemblance between SSMs and transformers does run deeper: the recurrence at the core of SSMs aggregates tokens just as attention does. Numerical experiments on a pretrained Mamba-2 model point to the output gate as the component that regulates the extent of this consensus, preventing the tokens from reaching it in full.

Acquiring telicity, which is the distinction between bounded (e.g., ate an apple) and unbounded (e.g., ate apples) events, requires first language (L1) learners to map surface-level and semantic cues to abstract event structures, but the computational trajectory of this mapping is not well understood. We introduce a Difference in Surprisal method that uses GPT2 token surprisal over paired temporal adverbial diagnostics (in an hour versus for an hour) to automatically label telicity across English CHILDES corpora, validated against expert linguist judgments. Using these labels, we train diagnostic logistic regression classifiers on 12 syntactic and lexical semantic features to compare how child speech and child-directed speech encode telicity. The two models diverge: the child model reaches near perfect accuracy through a single deterministic cue, the presence of a post-verbal determiner, while the adult model relies more heavily on verb class and other lexical semantic features, with the determiner cue neutralized. This trajectory supports Syntactic Bootstrapping: learners first exploit high-frequency structural cues as a scaffold to bootstrap, before developing fully compositional, verb-based event structures.