Papers

Few AI chatbots have transitioned the 'last mile' from research prototypes to clinical practice, partly because mapping conversational design decisions to regulatory requirements is complicated. This paper presents a design-oriented regulatory analysis of a chatbot supporting stroke survivors. Using Design Space Analysis operationalised through the QOC (Questions, Options, Criteria) framework, HCI researchers and legal scholars assess how variations in three conversational functions - health Q&A, recovery monitoring, and motivating medical adherence - influence qualification and classification under the European Medical Device Regulation (MDR), General Data Protection Regulation (GDPR), and AI Act (AIA). Our findings show that seemingly minor design decisions - such as role, actionability, and proactivity - influence regulatory obligations. The paper contributes insights into ten conversational design considerations influencing regulatory interpretation and demonstrates how QOC can support integrating legal considerations into human-centred design workflows. Finally, we suggest compliance-aware pathways to market, while reflecting on whether all chatbots should pursue the 'last mile'.

Lifelong (continual) anomaly detection systems must learn a sequence of tasks without forgetting how to recognize anomaly types seen earlier – a property known as backward transfer. A common way to try to improve backward transfer is to reshape a model's internal representation around learnable concept centroids, typically coupled with a new way of scoring anomalies based on distance to those centroids. Because these two design choices are usually introduced together, it is unclear which one is actually responsible for any performance gain. In this paper, we unbundle the two mechanisms within a VAE-based lifelong anomaly detector and study them separately. We find that used purely as a training-time regularizer – while leaving anomaly scoring unchanged – latent cluster separation can improve detection quality and backward transfer, though whether it outperforms the alternative scoring approach depends on the dataset's underlying representation. Switching to distance-based scoring is instead where most of the instability commonly associated with this class of method actually originates; we identify and explain the underlying mechanism. This benefit depends on the dataset's underlying representation, and we report this as practical guidance for choosing between the two design choices, rather than as a list of caveats.

Accuracy and outcome data are often treated as the main determinants of trust in automation. We test this assumption against an account in which clinicians evaluate artificial intelligence (AI) using social-cognitive cues. We introduce SPOT (Synergistic Process-Outcome Trust), a Bayesian framework that models trust as inference over latent agent properties from outcome and process signals. In a within-subjects vignette study, 275 Spanish healthcare professionals evaluated 16 clinical AI scenarios varying three outcome signals (accuracy, intervention timing, prognosis) and five process signals (interface familiarity, explainability, team consensus, supervision, personalization). Ratings were analysed with linear mixed models and participant-wise cross-validation. Process signals dominated trust formation. Interface familiarity (beta = 0.71), team consensus (beta = 0.64), and explainability (beta = 0.63) produced large effects, with p values below .001, whereas stated accuracy produced a small effect (beta = 0.10). Process signals accounted for 80.3 percent of the total absolute effect magnitude and outcome signals for 19.7 percent. Trust baselines were highly idiosyncratic and poorly recovered from demographics (R-squared = 0.06 under group-wise cross-validation). Clinicians appear to treat AI systems as agents whose trustworthiness requires social validation as well as technical verification. Building trustworthy clinical AI therefore involves social, institutional, and technical factors.

Domain-specific tabular and transaction foundation models are increasingly proposed as drop-in representation learners for financial applications, but their value relative to off-the-shelf text embedding models for retrieval-augmented generation (RAG) over structured data remains underexamined. We construct a deterministic, template-generated question-answering benchmark of 1,026 questions with retrieval-level gold-document annotations over IBM's synthetic Anti-Money-Laundering (AML) transaction dataset, and compare NVIDIA's 29M-parameter Transaction Foundation Model (TFM) against five general-purpose and financial-domain text embedding models under a size- and data-matched retrieval setup, as well as against a hand-aggregated summary-text baseline. TFM substantially outperforms every size-matched generic embedder on both retrieval (Recall@5: 0.81 vs. 0.25-0.39) and end-to-end answer accuracy (66.7% vs. 27.5%). However, a simple pipeline that embeds a human-engineered aggregated summary with a generic sentence encoder matches or exceeds TFM overall, and dominates it on several low-cardinality entity categories (bank, currency, payment format). This suggests that the benefit of a transaction-native foundation model over generic text embeddings is category-dependent rather than uniform, and that careful feature aggregation can substitute for domain-specific pretraining in several practically relevant regimes. We release the benchmark generation and evaluation code.

The variational graph autoencoder (VGAE) regularizes its posterior toward the prior with the Kullback-Leibler divergence, a choice inherited from the variational autoencoder rather than argued for. We introduce the generalized graph variational autoencoder (GGVA), which replaces that term with any member of the Rényi-Tsallis family of order $q$ while leaving every other part of the model untouched. Both members admit closed forms for diagonal Gaussians and both recover the KL exactly as $q \to 1$, so the VGAE is the $q=1$ arm of our own model rather than a separate baseline, and any measured difference is attributable to a single scalar. Our analysis identifies boundedness, not the order, as the operative property: for $q<1$ the Tsallis divergence is bounded above by $1/(1-q)$, independently of the latent width, whereas the KL and the Rényi divergence of the same order are unbounded. On ten graphs spanning three synthetic families, a social network, three citation networks, a connectome, a power grid and a road network, $q$ moves the retained posterior information by up to $49\times$ relative to the VGAE, while the Rényi arm at the same order stays within $1.02$-$1.30\times$ of it on all six larger real graphs (isolating the bound as the cause). The retained information is usable: probing the frozen embedding for node class, a label absent from the objective, gives GGVA up to $+0.14$ macro-F1 over the VGAE on CiteSeer, with the Rényi control again tracking the VGAE. We also report what the design was built to expose: none of this reaches held-out link-prediction accuracy on any of the six larger real graphs, and boundedness delays posterior collapse rather than preventing it.

Agent skills bundle procedural guidance with execution conditions governing authority, user consent, and live environment state. When model capabilities advance, maintainers prune instructions that appear redundant on authorized benchmark tasks. However, authorized maintenance tests can leave dormant safety conditions untested. This mismatch creates an unmeasured support gap over physical and privacy-sensitive effects. We introduce matched authority counterfactuals that hold the requested action, tool parameters, and intended effect fixed while systematically varying a single governing predicate. We formalize this evaluation via a two-gate retirement certificate requiring a candidate reduction to preserve authorized utility within a declared margin while producing zero unauthorized protected effects. In controlled experiments spanning four frontier and local model configurations across twelve skill bundles (2,592 evaluation cells), task-certified reductions remove over 94% of skill clauses and preserve authorized completion, yet produce unauthorized protected effects in every bundle. Boundary enforcement eliminates protected effects on the declared audit but fails the utility gate for one configuration. One bounded combined protocol passes both gates across all four configurations, with zero utility headroom. An end-to-end check on one read-only Home Assistant camera chain verifies proposal, decision, and effect measurements on a real device. These results demonstrate that while task benchmarks can justify retiring procedural guidance, retirement decisions require explicitly auditing the authority contracts governing physical actions.

Referring segmentation in overhead imagery is inherently relational: a query may ask for the buildings north of the road or the pond closest to a residential area, so the correct referent can contain one object, several objects, or none. Existing benchmarks mainly score mask overlap, which cannot verify whether a model actually resolved the stated spatial relation. We introduce GeoRefer-Bench, a benchmark for verifiable geospatial referring segmentation. Each query is represented by an executable logical form over a metric scene graph, and predictions are evaluated with Exact Query Success (EQS), which is satisfied only when the returned instance set exactly matches the set denoted by the query. GeoRefer-Bench contains 700 whole 2048x2048 UAV scenes (2.94 Gpx) at 12.5 and 25 cm ground sampling distance, 26,217 instances, 142,796 spatial relations, and 20,916 executable queries spanning five reasoning levels. It further includes three paraphrases per query, 24.0% unanswerable queries, 2,477 counterfactual pairs, and five leakage-controlled evaluation splits. An independent audit re-derives object geometry, mask ownership, relation values, query execution, and split provenance, finding zero issues across all 700 scenes. Relation-blind strategies can retain non-trivial mIoU while achieving at most 22.7 EQS overall, showing that overlap alone does not certify relational grounding. Across fifteen current models, the strongest reaches 74.1 EQS but drops from 98.9 at level 1 to 60.5 at level 5, while ten models score below 5 EQS on two-hop queries. GeoRefer-Bench turns geospatial referring segmentation from mask matching into verifiable reference resolution.

Chest radiographs are routinely used to verify the position of catheters, tubes, and other support devices. Existing image models often return labels or segmentations, while report-processing systems structure text without access to image geometry. We present an uncertainty-aware clinical knowledge graph that represents device instances, tip estimates, placement assessments, provenance, report events, and temporal links as separate but connected evidence. We evaluate the implemented visual graph layer using saved predictions from the complete RANZCR CLiP test archive, comprising 30,083 studies from 3,255 patients across five non-overlapping outer folds. The graph builder materializes 914,632 B7 evidence nodes and 884,549 typed relationships. All 118,647 B7 predicted-device nodes retain tip covariance, placement probabilities, fragment provenance, and fragment counts, whereas the direct B2 baseline retains none of these fields. We further define typed data contracts, uncertainty representations, abstention rules, report-image grounding, and longitudinal query mechanisms for extending the graph to report-bearing cohorts. The reported graph-materialization analysis is post-hoc descriptive and does not establish report grounding, longitudinal performance, or clinical utility. It demonstrates a reproducible foundation for evidence-preserving AI reasoning over chest X-ray device assessments.

Real conversations between fraudsters and their targets are among the most informative artifacts for studying telephone scams, yet also the scarcest: passive honeypots overwhelmingly capture automated messages and hang-ups, large-scale studies characterize call metadata rather than dialogue, and manual scam-baiting does not scale. We present a dataset of real scam-call conversations collected by an active voice-agent honeypot. Dedicated numbers are seeded into the lead-generation channels fraud operations harvest; inbound callers are answered by a low-latency conversational agent that adopts a plausible target persona and sustains the interaction while every call is recorded, transcribed, and automatically labeled. Over an initial 53-day window we captured 10,015 inbound scam and spam calls (6,601 with two or more turns): roughly 895 hours of audio and 328,869 transcribed turns from 5,665 distinct originating numbers. Under a holistic classifier the substantive calls are predominantly predatory-but-legal lead generation ("spam", about three in five), while about one in seven is an outright "scam" (949 in this snapshot). Each call carries a turn-level transcript, three-channel audio, per-turn latency telemetry, and layers of automatic labels, including a holistic scam/spam/legitimate judgment corroborated by independent human review (75% agreement on the binary decision). We describe the collection system, the record structure, and technical validation of the corpus's realism and label quality, including that the agent is recognized as non-human in only about 5% of engaged calls. We also benchmark established scam-detection methods, where detectors trained on published synthetic dialogue collapse in precision on real traffic.

Automated microscopy could widen access to malaria diagnosis in low-resource settings, but the deadliest species, P. falciparum, presents in its early ring stage as an object only a few tens of pixels wide. Such tiny targets are systematically under-detected: overlap-based label assignment starves them of positive samples, and overlap-based box regression gives weak gradients at their scale. The Normalized Gaussian Wasserstein Distance (NWD) repairs both effects, but applied uniformly across a slide that also holds objects three to four times larger it loosens their supervision and erodes their localisation, so overall accuracy can fall even as the tiny class improves. We present CoSWA-YOLOv12, a compact YOLOv12 instance-segmentation detector whose core Cooperative Scale-adaptive Wasserstein Assignment routes the Wasserstein treatment to an object in inverse proportion to its size, tapering back to standard assignment for larger species. Two further components support it: a wavelet detail residual, and a min-max Gaussian regression loss (M2-NWD). All three additions are transfer-safe: each reproduces the standard pretrained model exactly at initialisation, so public pretrained weights load without any loss of accuracy. On a five-class Rwandan thick-smear dataset, CoSWA-YOLOv12 raises P. falciparum recall from 0.63 to 0.74 and mAP@50 from 0.73 to 0.81 (mask), cuts missed P. falciparum from 38% to 15%, and improves strict-localisation mAP@50-95 on all five classes for both detection and segmentation, while a 2x2 ablation shows the scale gate and the regression loss are synergistic.

A representation-adaptive kernel class produces, on a fixed sample, a union of reproducing-kernel Hilbert-space ellipsoids rather than one ellipsoid. We introduce the minimum-trace common covariance that dominates the unrestricted empirical union generated by Brownian kernel ladders and develop its statistical, approximation-theoretic, and computational consequences. The covariance value admits exact formulations through absolutely two-summing operators and covariance-dominated multipliers, and it yields a universal Gaussian-complexity bound. A closed last-layer Dirac-trace reduction and a signed Brownian threshold representation convert the generic covariance problem into threshold, graph-coarea, and effective-resistance geometry. These tools give deterministic depth laws, conditional Gaussian reverses, random-design and perturbation transfers, and an exact empirical Kolmogorov-width formula whose leading covariance eigenspaces approximate the complete adaptive ball simultaneously. Finite contact, active semidefinite programs, verified separation, and a convex resistance-design relaxation provide complementary lower and upper certificates. A finite covariance-indexed Brownian path on frozen representations illustrates the distinction between successful covariance certification and predictive selection: all reported path certificates succeed, whereas the locked predictive study misses one predeclared aggregate criterion. The paper thereby identifies one finite-dimensional covariance object linking unrestricted kernel adaptation, Gaussian geometry, common subspaces, and certifiable computation.

We construct and analyze structure-preserving numerical schemes for a two-stage local–nonlocal dispersal system. On general bounded connected smooth habitats, we prove fixed-$\delta$ spatial consistency. We introduce a semi-implicit scheme that is uniquely solvable for every $\Delta t>0$, preserves nonnegativity, and exactly retains the sign of the semidiscrete spectral threshold. A negative threshold yields geometric extinction of the numerical solution, while a positive threshold makes the zero state of the numerical scheme linearly unstable. On rectangular habitats, we construct a reflected Cartesian discretization that is asymptotically compatible with the Neumann local limit, with convergence estimates uniform with respect to the ratio between mesh size and interaction scale. Numerical experiments illustrate the structural, threshold, and local-limit behavior.

Histology and transcriptomics provide complementary views of tissue biology, capturing spatial morphology and molecular activity, respectively. Pathology foundation models (PFMs) learn rich morphological representations from H&E images, yet strong downstream performance alone does not establish whether these representations encode biologically meaningful and robust molecular information. We present a **training-free framework for auditing biological alignment in frozen PFMs** using spatially paired histology and transcriptomics from HEST-1k, evaluated on **240 samples spanning three organs**. Multiple frozen PFMs are used to extract H&E representations, while gene expression is aggregated into biologically interpretable pathway-level programs. We use TabPFN as a pretrained probe to quantify the extent to which these molecular programs can be decoded from frozen image representations without task-specific gradient updates. Beyond predictive performance, our audit examines whether pathway decodability generalizes across tissue sections, patient groups, and tissue types; whether representations exhibit section-level or other shortcut dependencies; and whether predictions remain stable under small image perturbations and context resampling. This multi-tissue evaluation distinguishes molecular programs that are consistently encoded from those that are tissue-specific, unstable, or shortcut-sensitive. Our framework therefore provides a systematic approach for assessing not only **what biological information pathology foundation models encode, but also how reliably that information persists under clinically relevant sources of variation**.

Tool-using language-model agents increasingly mutate schedulers, data pipelines, object stores, and access-control systems. Between an agent's read and its commit, external state can change, but not every change makes the commit unsafe. We separate invalidating races, which break a declared safety predicate, from predicate-preserving and irrelevant races, and ask how precisely runtime guards distinguish them. Our deterministic simulator separates visible from authoritative state and injects five non-atomic failure mechanisms across 16 infrastructure tasks in four domains; frozen agent proposals are replayed counterfactually under every controller without an LLM judge. We evaluate three commit-time guard granularities (global epoch, read-set version, semantic commit predicate), multi-level verification, and model-side gates on three locally hosted quantized model families (Qwen3-4B, Phi-4-mini, Gemma4-8B; 3,456 trajectories on one GPU). All three guards eliminate unsafe commits, but their availability differs sharply: freshness-based guards needlessly block 92-95% of benign races, forfeiting up to 43% of safe task completions, while the complete predicate guard blocks none. That precision is contract-dependent: deleting a single declared clause converts exactly its fault family into unsafe commits (up to 7.9%). Model-side signals do not substitute: verbal confidence is miscalibrated (ECE approximately 0.37), action agreement matches a random gate, a cautionary prompt leaves the direct unsafe rate essentially unchanged, and after a freshness-guard block agents re-commit unsafely from refreshed but still-incomplete reads. Under degraded telemetry a hidden concurrent mutation remains observationally clean, bounding every selective policy. Precise runtime enforcement therefore requires semantic contracts, not freshness heuristics or model self-assessment.

Multitask predictors apply task-specific functions to a shared representation. Penalizing the representation and task functions separately can assign different costs to equivalent predictors and charge for directions the tasks do not use. We study the trace norm of the complete task predictors in a fixed input reproducing kernel Hilbert space, weighted by task sample sizes. For a finite-dimensional intermediate function space, this norm is the infimum of the product of the representation operator norm and task trace norm over invertible coordinate changes. Task inner products determine this norm; standard duality gives the exact empirical Rademacher complexity of its norm ball. We estimate the feature Gram matrix used in fitting. Under bounded leverage, one estimate controls the penalty for every task head and shared linear combination of fixed features, even when the Gram is singular and the coefficients are chosen after estimation. This yields comparisons of penalized empirical objectives and a generalization bound under an estimated norm constraint, with the stated loss and sampling assumptions. Controlled experiments show that signal alignment can favor either penalty. A SARCOS comparison finds close agreement between exact and estimated penalties on overlapping training and evaluation observations.

Mental health disorders affect hundreds of millions of people around the world, yet access to professional counseling remains severely limited. AI-powered dialogue systems offer a scalable alternative, but existing models face two fundamental challenges. First, they lack the bidirectional understanding needed to capture the layered nature of emotional expression, particularly in cases of progressive disclosure, where clients often present symptoms at the surface-level while concealing deeper trauma. Autoregressive (AR) models process information sequentially and cannot revise early interpretations when new evidence emerges later in the conversation. Second, they fail to effectively incorporate the relational knowledge that underlies clinical reasoning. In this paper, we propose BiGraph-Diffuse, the first large-scale diffusion language model tailored for the counseling domain. We further introduce BiGraph-RAG, a relation-free graph-structured retrieval strategy that relies only on lightweight entity extraction and semantic linking. This design preserves inferential pathways from observable symptoms to potential underlying causes, while incurring zero LLM token cost during indexing. Importantly, these two modules are not merely combined but mutually reinforcing. The diffusion model provides a holistic bidirectional context, enabling the system to defer premature judgments during progressive disclosure. Meanwhile, graph-based retrieval captures the structured interconnections of clinical knowledge. Extensive experiments demonstrate the effectiveness of BiGraph-Diffuse, and we further provide a solid theoretical analysis to support its design.

Reinforcement learning (RL) and on-policy distillation (OPD) are two representative paradigms for improving large language model reasoning. However, when no correct trajectory is sampled, RL lacks a positive correctness signal, while OPD remains constrained by the reasoning trajectories reachable under the student's on-policy distribution. Therefore, we propose CataOPD, where the teacher acts as a catalyst rather than a target, expanding reachability while internalizing verified student-produced trajectories into a catalyst-free policy. Self-Rescue Routing uses empirically all-failed groups as routing signals rather than teacher-intervention triggers, first seeking correct trajectories through additional on-policy self-sampling. For problems unresolved after self-rescue, Catalytic-Guided Self-Resolution uses catalytic guidance to elicit a verified student-produced trajectory in the guided student distribution. Barrier-Weighted Internalization weights tokens by guided-to-unguided log-probability gaps, focusing updates on decisive tokens difficult without guidance. Experimental results show that CataOPD outperforms current baselines, extends independent student reasoning to still-unrecovered problems, and improves out-of-distribution generalization under catalyst-free inference. Our project is available at https://github.com/QwenQKing/CataOPD.

Existing methods for improving text-to-image generation quality have progressed from generator fine-tuning and prompt optimization to reinforcement learning with multi-turn visual feedback. However, existing strategies are deeply coupled with specific generators and tasks, and the learned capabilities are difficult to generalize into a universal quality optimization policy. Therefore, we propose AdaPilot, which learns a scene-adaptive, cross-generator transferable quality optimization policy by formulating multi-turn image generation as a Markov Decision Process (MDP) and optimizing it via end-to-end reinforcement learning. Specifically, AdaPilot decouples the policy from generator internals to enable cross-generator transfer, introduces scene-aware rewards that adaptively align quality assessment dimensions with task semantics, and employs process-level rewards to model the evolution trajectory of image quality. Experimental results show AdaPilot outperforms baselines in generation quality and generalization. Separate cross-generator evaluations further show that a single policy transfers zero-shot to unseen generators while maintaining positive average gains across all evaluated generators. Our project is available at https://github.com/QwenQing/Ada_pilot.

Fairness-of-exposure treats algorithmic attention as a good to be distributed equitably. But when an autonomous agent initiates contact, attention is signed: it delivers value to a willing receiver and imposes a burden on an unwilling one. We formalize routing under signed exposure and show that a fair distribution of attention need not be a fair distribution of unwanted attention. Our central result is an incompatibility: within signed-exposure routing, exposure parity (equal contact rates across groups) and burden parity (equal unwanted-contact rates) generically cannot hold at once, and the two are separated by a band that widens as routing grows more selective. A second result shows measurement error is itself a fairness mechanism: group-differential noise in receptivity scores simultaneously inflates a group's exposure and degrades whom it selects, so an apparent exposure-fairness gain is a hidden burden transfer. Calibrating to a public dating-platform survey (n=2,499) that, to our knowledge, uniquely measures receive-side receptivity to conversational agents, we find exposure parity costs only 0.2–2.3% of yield yet moves the per-capita burden ratio to 1.7 times: the tension is between fairness notions, not between fairness and efficiency. Finally, the burden-parity policy is computable by bisection and learnable online: a plug-in learner recovers it at a $2.3\%$ empirical regret premium. The operative design choice in signed-exposure markets is not efficiency versus fairness but which fairness.

Machine Translation (MT) for low-resource languages remains far behind that of high-resource languages, and the gap is widest in specialised domains, where parallel data is scarce or entirely absent. We present EnSiTa, a trilingual multi-domain parallel dataset and benchmark for English, Sinhala and Tamil. EnSiTa provides human post-edited training data for seven domains, plus manually translated test sets for those and one additional domain, all produced by professional translators under a multi-year, rigorously quality-controlled process. Using this dataset, we conduct an extensive study of domain-specific MT for all six language directions, fine-tuning a from-scratch Transformer, a pre-trained translation model (NLLB-600M), and decoder-only LLMs (Gemma 3 family, 1B-12B, and TranslateGemma) across training-data sizes, model scales, and in-domain, cross-domain, multilingual and multi-domain settings. To the best of our knowledge, this is the most extensive systematically documented multi-domain parallel data creation and benchmarking effort for low-resource MT. Our data and models will be publicly released.

Implicit neural representations (INRs) have emerged as a promising paradigm for video compression, providing compact neural representations with flexible spatial and temporal reconstruction. Hierarchical grid-based architectures such as HiNeRV achieve strong rate–distortion performance, but require extensive per-video optimization, resulting in high encoding costs. To address this limitation, we propose ScoutNeRV, a content-adaptive initialization framework for accelerating the optimization of hierarchical video INRs. ScoutNeRV employs a lightweight, offline-trained scout network that analyzes a small number of sampled frames and selects a suitable pre-trained expert from a memory bank through hard routing. The hierarchical grid and decoder parameters of the selected expert are then transferred to initialize the target HiNeRV model before video-specific fine-tuning. On the unseen ReadySetGo sequence, ScoutNeRV achieves an initial PSNR of $34.95$~dB, compared with $13.70$~dB for standard initialization, corresponding to a $21.25$~dB improvement before fine-tuning. After only 37 epochs, ScoutNeRV reaches $36.92$~dB and remains within $0.42$–$0.80$~dB of the 300-epoch HiNeRV baseline across the evaluated rate–distortion configurations. Furthermore, the proposed initialization achieves a $9.25\times$ wall-clock speedup in the reported runtime experiment. These results demonstrate that content-aware expert initialization can substantially reduce the optimization cost of hierarchical video INRs while retaining competitive reconstruction and compression performance. The code is available at https://github.com/nasserdeveloper/ScoutNeRV.

While Vision-Language Models (VLMs) demonstrate strong capabilities, they continue to suffer from a critical limitation: insufficient fine-grained visual perception, which fundamentally limits their multimodal understanding. We attribute this bottleneck to text-dominant optimization biases during pre-training, which encourage the model to overlook fine-grained visual details, thereby limiting the capability of multimodal understanding. We investigate that overcoming this bottleneck requires two key elements: (1) a unified token space paradigm that ensures stable training dynamics, and (2) a modality-aligned dense visual supervision signal enriched with both structural granularity and semantic information to capture critical visual representations. Based on these insights, we propose VIVAS, a framework built upon the unified token space paradigm, which introduces a dense-structural-semantic vision tokenizer, which expands the textual vocabulary into a unified vision-language vocabulary by incorporating a visual vocabulary. During pretraining, VIVAS performs vision-language unified autoregressive supervision over both visual details and linguistic content, thereby enhancing visual perception to improve multimodal understanding. Trained end-to-end on 12.4T tokens, VIVAS achieves state-of-the-art performance across 7 tasks and 39 multimodal benchmarks.

Automated Market Makers enable decentralized trading but systematically expose liquidity providers to impermanent loss through arbitrage-driven rebalancing. While dynamic fee mechanisms offer a promising mitigation strategy, existing approaches remain largely reactive, adjusting costs based on historical signals rather than explicitly linking them to the structural risk imposed by individual trades. To address this limitation, we propose a novel fee formation framework built on three core innovations. First, we introduce a coupled market maker architecture in which fee dynamics are governed by a secondary invariant, allowing liquidity state and transaction costs to evolve jointly. Second, we develop an impermanent-loss trimming fee model that adaptively increases transaction costs for trades exceeding the liquidity providers' profitable region, effectively offsetting losses from large arbitrage executions while preserving baseline fees for smaller transactions. Third, we establish a unified evaluation methodology using performance profiles to systematically compare fee algorithms across diverse market conditions. By extending a heterogeneous trader model to derive optimal arbitrage strategies under state-dependent fees, we conduct extensive simulations on historical data spanning four distinct market regimes and three token pair categories. Our results demonstrate that the proposed fee enhancements improve liquidity provider yields by 6–24% in volatile markets and up to 119% in calm regimes, while maintaining uninformed user participation and reducing informed arbitrage profitability by 3–10%. Performance profile analysis confirms that ILT-enhanced algorithms dominate baseline counterparts across 60–75% of test scenarios.

Despite remarkable advancements in multimodal large language models (MLLMs), their fine-grained visual understanding is constrained by a primary reliance on sparse textual supervision. Existing efforts to introduce visual supervision typically do so during post-training, when visual representations have already been largely fixed, causing such signals to act mainly as auxiliary constraints rather than as a primary force for shaping perceptual features. In this paper, we aim to fundamentally reshape the model's perceptual backbone by incorporating vision supervision directly into the pre-training stage. We observe that pixel-level image patches and textual tokens naturally coexist in a shared, raw high-dimensional space characterized by an inherent input symmetry. Leveraging this insight, we propose UVU, a novel vision-language unified autoregressive framework that eschews vector quantization. It uniquely employs continuous visual encoding for lossless representation of visual inputs and proposes a large-scale iterative hierarchical clustering algorithm to construct a pixel-level visual codebook, thereby extending the vocabulary for unified supervision and enabling autoregressive generation of pixel-level image tokens alongside textual tokens. UVU effectively synergizes pixel-level visual perception with semantic-level visual understanding, internalizing visual reconstruction capabilities and unlocking the facilitative role of visual supervision in enhancing understanding in the pre-training stage. Extensive experiments across multiple tasks demonstrate that MLLMs are capable of achieving superior multimodal understanding performance under the supervised learning paradigm of UVU.

Social dynamical networks significantly influence contemporary digital landscapes, affecting realms from social activism to public policy formulation. This paper investigates the use of multi-agent systems (MAS) to monitor and analyze these networks. Firstly, we propose a single-time-scale distributed inference model designed to effectively manage challenges such as latency and agent failure. Secondly, we provide sufficient conditions that ensure the stability of the proposed scheme. Notably, the observer gain design remains effective regardless of time delays. Thirdly, we develop a computationally efficient recovery mechanism for agent failures that relies on employing graph-theoretic approaches to restore network observability by replacing failed agents (due to sensing failure or unbounded delays, i.e., packet drops) by implementing computationally efficient graph-theoretic methods to assign observationally equivalent agent counterparts. Lastly, we illustrate the proposed scheme through a pedagogical example and real-world network applications.

Human-AI Teaming (HAT) reviews often group studies by labels such as advisor, teammate, or coordinator. Yet the same label can describe one person taking AI advice, several people coordinating around AI, or a workflow that distributes authority and responsibility. Pooling these studies can therefore change the human unit behind a claim. We examine how full-text evidence changes the set of studies behind a claim. We audited 86 full texts purposively selected from a 419-record title/abstract map. We find that full-text reading changed core membership for 40 records: 36 of 74 apparent core candidates moved out, while 4 of 12 boundary candidates moved in. Team vocabulary did not reliably identify the social unit: 14 of 27 human-AI dyads and 20 of 23 multi-human peer teams used team or collaboration terms. Only 20 of 86 papers specified who could see AI output. Four blinded language-model runs unanimously labeled 53 screening cases and 59 arrangements, yet 32% and 34% of those consensus decisions differed from the full-text labels. These results identify claim-denominator drift as a synthesis problem in HAT research. We contribute a full-text audit centered on human arrangements and a claim-pooling checkpoint for deciding when evidence about trust, coordination, performance, efficiency, and accountability can be compared.

Machine learning learns functions: prompt to response, image to caption. What these functions are mathematically remains hard to say. We present a method to approximate these kinds of transformations using techniques from dynamical systems that fall under the umbrella of Koopmanism. We introduce the use of composition operators, which generalize the Koopman operator and, crucially, can map between distinct spaces, motivating the perspective that LLM transformations are rectangular infinite-dimensional operators. This formalism reveals useful structure: under natural assumptions on the prompt and response distributions, the LLM operator is an isometry, and misalignment between learned representations manifests as spectral pollution of its finite sections. We then outline a method of constructing finite-dimensional approximations of an LLM operator, and demonstrate how the singular value spectrum can be used to compare tasks and models.

Kubernetes scheduler plugins that score candidate nodes are, in production, hand-tuned heuristics. We ask whether a learned scoring function - trained on real placement decisions from a production cluster trace - can match or exceed these heuristics, and if not, why. We implement an external, HTTP-backed scoring plugin for a widely used scheduler simulator and evaluate two learned models (a Random Forest over engineered features, and a graph neural network encoder over per-job task-dependency graphs) trained on a large-scale production cluster trace. Under standard regression fit (R^2), both models improve modestly but monotonically across four feature-engineering iterations, reaching R^2 of about 0.042. However, on the metric that actually matters for scheduling - Top-1 ranking accuracy, whether the model scores the machine the production scheduler actually chose highest - both learned models are outperformed by a trivial single-feature heuristic (rank by free CPU: 74-84% vs. 65-66% for either model). We show this gap is best explained by objective mismatch: both models were trained with pointwise regression (MSE) rather than a ranking-specific objective, echoing a long-standing distinction in the learning-to-rank literature. This parallels prior evidence that RL-trained schedulers can outperform heuristics when the training objective is aligned with the deployment task, suggesting objective misalignment, not architecture, is the primary obstacle here. We further report an ablation of the occupancy reconstruction required to make offline trace data usable (naive features yield R^2 near 0), a controlled comparison isolating feature richness and data volume between the two model families, and a sensitivity analysis of inference latency and serving-container memory constraints. Code, data pipelines, and experiment scripts are released for reproducibility.

Automated seizure detection from scalp electroencephalography (EEG) is difficult because seizure morphology varies among patients and seizure samples are substantially outnumbered by non-seizure samples. This paper presents an attention-enhanced dual-branch network that jointly learns time–frequency and temporal representations from the same EEG segment. A continuous wavelet transform converts each segment into a scalogram processed by an ImageNet-pretrained ConvNeXt-Tiny backbone and squeeze-and-excitation attention. In parallel, a bidirectional long short-term memory network followed by multi-head self-attention models the raw signal. The two feature vectors are concatenated and classified by a weighted multilayer perceptron. Experiments use 14 subjects from the CHB-MIT scalp EEG database with subject-wise partitioning performed before overlapping segmentation. The model obtains $97.88\%$ accuracy and $97.51\%$ F1-score over ten across-subject splits, and $97.51\%$ accuracy, $96.59\%$ F1-score, and $98.02\%$ area under the ROC curve under 14-fold leave-one-subject-out validation. Removing temporal attention causes the largest ablation loss. The model requires 28.26 million parameters and 4.56 GFLOPs, with a measured network-only inference latency of 4.64 ms per segment.

Automatic modulation classification (AMC) of received radio signals is prudent for further signal processing tasks such as communication monitoring, cognitive radio operation, and interference mitigation in the electromagnetic spectrum. Traditional methods often rely on handcrafted features and struggle under complex channel conditions, whereas deep learning (DL) architectures can learn discriminative representations directly from raw signals. In this work, we propose HFEMCNet, a novel compact hybrid frequency-enriched multi-channel network that jointly exploits spatiotemporal dependencies and frequency domain information derived via Fast Fourier Transform (FFT). HFEMCNet integrates raw IQ samples with hierarchical spectral features to produce a rich signal representation, which is processed by convolutional layers for spatial feature extraction and Long Short-Term Memory (LSTM) units for temporal modeling. Experimental evaluation on benchmark datasets RML2016.10a, RML2016.10b, and over-the-air RML2018.01a demonstrates that HFEMCNet significantly outperform contemporary state-of-theart DL models in classification accuracy while achieving a reduced parameter count, smaller memory footprint, and lower tail-latency, making it suitable for real-time deployment on resource-constrained platforms. These results highlight the effectiveness of combining hybrid architectures with frequencydomain enrichment for robust and efficient AMC.