Papers

Identifying the bug-inducing commit (BIC) is a fundamental step in regression debugging and a key input to emerging BIC-aware fault-localization pipelines. In practice, BICs are commonly obtained with bisection. Standard bisection selects the median commit of the remaining good-bad interval, thereby balancing commit count. This strategy is optimal under the assumption that each commit is equally likely to be the BIC. This paper shows that this assumption does not match real-world BIC histories. We construct a dataset of 8,172 bug reports from GCC, the Linux kernel, and MariaDB. We find a strong temporal skew: across the studied systems, 50% of BICs lie within the most recent 0.69% of the report-time commit history. Motivated by this observation, we introduce BISCEPTER, a probability-driven bisection approach that uses historical BIC latency as a lightweight prior. Instead of selecting pivots that split the number of remaining commits, BISCEPTER selects weighted-median pivots that split estimated BIC probability mass, while preserving the same good-bad oracle and interface as standard bisection. We evaluate BISCEPTER on three large-scale systems. The evaluation results show that BISCEPTER reduces bisection iterations by 25.75% on average (up to 55.55%) compared with standard median bisection, while improving over the baseline in 91.26% of test cases. Robustness experiments further demonstrate that the benefit remains stable under noisy historical data. We expect that our research can effectively save effort in debugging software in practice and, more broadly, benefit future software engineering research by bringing insights about BIC distribution.

LLM agents often fail mid-task due to invalid tool calls, repeated actions, or poorly grounded reasoning, and learning from these failures is a path to reliability. We find that how failure knowledge reaches the agent matters as much as what it contains. Failure lessons are conditional: kept in the agent's context, they misfire when their failure is absent, and removing them from an evolving playbook improves performance. Runtime interventions, in contrast, act only when a failure occurs but do not learn from their repairs. We argue that failure knowledge is conditional knowledge and should be conditionally exposed, and instantiate this principle in Sentry, a failure-management layer that runs alongside the agent. When Sentry detects a failure, it retrieves matching lessons from an external playbook to guide recovery, verifies without access to task rewards whether the agent recovered, and stores a new lesson only if it did; the full playbook never enters the agent's context. Across multiple agentic benchmarks, Sentry outperforms the strongest runtime-intervention baseline on every benchmark, by 37% on average, and the strongest context-evolution baseline by 39% on the two benchmarks where both are evaluated; combining Sentry with context evolution yields further gains. Learned lessons transfer to held-out tasks, and controlled experiments show that exposing the full playbook to the agent lowers performance even when relevant lessons remain available on demand.

Lower-limb exoskeletons have made considerable progress in reducing energy expenditure during walking. However, designing optimal assistance strategies remains challenging, particularly given inter-individual variability in anthropometry and biomechanics. This study explores energy-optimal lower-limb joint-assistance strategies using predictive simulations with musculoskeletal models. Subject-specific models of six able-bodied subjects, with BMI-based muscle strength scaling, were used in predictive simulations to generate gait at self-selected walking speeds. Ideal actuators were incorporated to simulate various combinations of joint assistance at peak levels of 25 Nm and 50 Nm to examine the effects of assistance on gait and metabolic savings. The effects of each assistance configuration were assessed through cost of transport (COT), joint kinematics, muscle activations, assistive torques, and joint-level power metrics to characterize the biomechanical and energetic impacts of different assistance strategies. At 50 Nm, combined H+K+A (hip-knee-ankle) assistance resulted in the greatest mean COT reduction of 48.50 +/- 4.55%, with individual reductions ranging from 42.22% to 54.65% across the subjects. Among single-joint conditions, assisting the hip was most effective, reducing COT by 31.77 +/- 5.21%; the knee and ankle produced smaller, comparable reductions (19.17% and 17.02%). H+A (hip-ankle) assistance (44.60 +/- 7.08%) emerged as the most effective two-joint assistive configuration. Increasing the torque bound increased positive assistive power primarily at the hip and ankle, while knee assistance showed little sensitivity and delivered positive power near pre-swing. These results support H+A assistance as an efficient two-actuator target, while identifying the knee's atypical pre-swing power strategy as a candidate for targeted experimental validation.

Real-world interaction networks are inherently dynamic: edges form and dissolve as node behavior shifts over time. Most snapshot-based contrastive methods encode temporal dependencies implicitly in encoder weights, without an explicit model of how node representations evolve, making them brittle under distribution shifts. We propose KAIROS (Koopman-Aligned Invariant Representations for Open Dynamic Systems), a self-supervised framework that embeds a differentiable Koopman operator within a dynamic graph contrastive learning loop to linearize temporal evolution in the learned embedding space. A dual-view encoder pairs raw node features with a graph-diffused structural view and is optimized with multi-granularity contrastive objectives across temporal windows. For anomaly detection, KAIROS uses the Koopman prediction residual together with temporal inconsistency and local neighborhood deviation to separate irregular behavior from predictable graph evolution. Evaluated on nine dynamic graph benchmarks, KAIROS achieves state-of-the-art anomaly detection results on all nine datasets, with gains of up to 23.15 ROC-AUC points over prior work, while remaining competitive for unsupervised node classification. These results show that explicit dynamics modeling provides a scalable and effective inductive bias for temporal graph representation learning.

Membership inference on large language models (LLMs) aims to determine whether a given text sample was included in an LLM's training data, without access to its training corpus. Despite recent progress, existing benchmarks suffer from three limitations: limited coverage of training stages, insufficient distributional alignment between members and non-members, and lack of rigorous filtering of non-members against the training corpus. To address these limitations, we propose OLMo-Detect, a multi-stage, confounder-controlled benchmark built upon the fully open OLMo 2 pipeline. OLMo-Detect spans pre-training, mid-training, and post-training, explicitly aligns members and non-members on three key axes, and rigorously filters non-members via infini-gram. To assess robustness to distribution shifts, we further introduce OLMo-Detect (Shifted), a variant where members are misaligned with non-members. We evaluate 15 unsupervised and 3 supervised membership inference attacks (MIAs) across the OLMo 2 family, finding that: (i) overall performance is limited: the best unsupervised and supervised MIAs both reach an AUC of only 0.68, and supervised MIAs degrade under cross-domain evaluation; (ii) MIA performance peaks at mid-training and is lower at pre-training and post-training, a pattern driven by data type rather than a stage effect: curated math data is far more detectable than other types; (iii) overall scores improve from 1B to 13B but plateau at 32B; and (iv) no unsupervised MIA is robust to distribution shifts, with AUCs shifting by up to 0.42. Finally, we find that our findings on OLMo 2 generalize to OLMo 3 and non-OLMo models.

Embodied agents interacting with the same physical process may maintain heterogeneous, asynchronous, and observer-relative representations. Rather than assuming that such representations should always be globally aligned, we investigate which distinctions among them must actually be resolved for coherent interaction. We formalize this question through relational sufficiency: an interaction-specific relational-evidence map defines the representational ambiguity left unresolved by comparison, and sufficiency holds when each residual ambiguity fiber lies within an interaction-equivalence class. When this condition fails, additional sensorimotor anchors must separate precisely the ambiguities that can change the interaction-relevant outcome. For smooth systems, we derive a local first-order lower bound on the number of inde- pendent scalar anchors required, given by the dimension of directions that are invisible to relational evidence but visible to the interaction-relevant map. We instantiate the theory in two variants of an asynchronous hand-off scenario that share the same observer agents, representations, transports, and relational evidence. Relative synchronization requires no additional anchor, whereas hand-off at an absolute time requires exactly one. The analysis there- fore provides a task-relative stopping condition for representational alignment: residual disagreement need not be eliminated when it is invariant with respect to the interaction.

Direct approaches to discrete parabolic Carleman estimates often rely on weighted identities tailored to the discrete operator. We prove an abstract transfer theorem that derives such estimates for fully discrete backward Euler schemes with conforming spatial reconstructions from a continuous Carleman inequality together with suitable weighted approximation and compatibility properties. For a class of time-independent positive self-adjoint operators, the key construction assigns to each discrete state a shifted continuous auxiliary problem whose discrete solution is exactly the prescribed state. Weighted space-time error estimates, uniform over the full discrete load space, then transfer the inequality while preserving the residual and observation terms of the original scheme. This separates the PDE-specific Carleman analysis from the numerical verification required for each discretization. We verify the hypotheses for the heat equation discretized by consistent-mass $P_1$ finite elements on locally graded meshes in two and three dimensions and by standard Cartesian finite differences in any fixed dimension. For the heat equation, the transferred estimate is valid for Carleman parameters up to the scale $\min\{h^{-4/5},(\delta t)^{-2/5}\}$. For bounded real-valued space-time potentials, the resulting estimates yield relaxed observability and approximate null controls with uniformly bounded cost and exponentially small terminal errors under independent spatial and temporal refinement; weak accumulation points of the reconstructed controls are continuous null controls.

Programmable logic controllers (PLCs) coordinate industrial equipment by reading sensor inputs and issuing control commands. Evaluating whether large language model (LLM)-generated PLC programs satisfy task requirements and safety constraints requires observing how their commands affect device and workpiece states. We introduce PLCWorld, a common closed-loop execution environment and benchmark that couples Structured Text (ST) execution with simulated plant responses and sensor feedback. Grounded in control relations identified in industrial PLC programs and engineering documentation, PLCWorld contains 100 synthetic tasks and 473 registered task-condition pairs across Motion Control and Material Handling, with difficulty defined by control-dependency scope. A common protocol reports Task Success and Safety Violation separately. Validation combines practitioner review, reference and alternative programs, targeted counterexamples, specification-evaluator alignment checks, and comparisons with independent ST runtimes. Reference and alternative programs satisfy their applicable cases, while all 542 targeted counterexamples activate their designated evaluator rules under at least one registered condition. Execution Gap relates submission-profile acceptance to subsequent task failure or observed Safety Violation. Across the constructed task groups, direct GPT-5.5 achieves 82.70% Task Success on Easy cases but 25.10% on Hard cases. Evaluations of six LLMs and four adapted generation-and-verification workflows further expose differences between completion, safety, and generation cost. Our code, simulation environment, benchmark tasks, and baseline implementations are publicly available at https://yunji0516.github.io/PLCWorld/.

Source-Free Domain Adaptation (SFDA) aims to adapt a source-pretrained model to an unlabeled target domain without access to the original source domain. While early single-model approaches rely on self-refinement, they are inherently susceptible to confirmation bias and struggle to correct their own systematic errors. To overcome this limitation, recent methods introduce Vision-Language (ViL) models as external knowledge sources. However, these approaches operate in a largely unidirectional paradigm, using the ViL model primarily to supervise the source-pretrained model. This overlooks a key structural property: the two models exhibit distinct failure modes – where one produces an incorrect prediction, the other may produce a correct one, creating a natural opportunity for mutual correction within the target domain. Yet, without ground-truth labels, identifying which model is correct on any given sample is non-trivial, and naively exchanging predictions risks propagating errors across models. To address this challenge, we propose SafeCut, a novel approach that leverages the cut statistic as a label-free measure of prediction reliability to gate cross-model supervision. Our approach dynamically controls both the direction and strength of supervision based on relative reliability, selectively amplifying true corrections while suppressing miscorrections on a per-sample basis. We further provide theoretical justification showing that this reliability-gated mechanism guarantees a net-positive correction signal. Extensive experiments across diverse SFDA benchmarks demonstrate that SafeCut achieves state-of-the-art performance, highlighting the effectiveness of safeguarding mutual correction in SFDA via cut statistics.

Unstructured discharge notes in Electronic Health Records (EHRs) often carry signal complementary to structured medical codes, holding patient-specific evidence that standardized cohort-level codes alone cannot capture. However, this evidence in notes is frequently buried in lengthy, noisy text that is not intentionally written with any specific clinical prediction in mind. Summarization is an obvious mitigation, but generic summaries, tuned for fluency rather than the outcome, routinely omit decisive evidence while retaining plausible but uninformative detail. To this end, we propose RASPER, a Reward-Aligned Summarizer for Prediction in EHR, that optimizes note summarization directly against the downstream clinical task. RASPER employs a tunable LLM-based summarizer to extract task-relevant evidence from discharge notes and trains it via reinforcement learning from prediction feedback, using a reward derived from the downstream predictor's loss. To ground the summarizer, a longitudinal encoder converts structured codes into soft prompts that incorporate each patient's clinical context into note summarization. By rewarding the quality of the resulting multimodal prediction, RASPER encourages the summarizer to retain patient-specific evidence that complements, rather than duplicates, information captured by structured codes. RASPER consistently outperforms strong baselines on both readmission prediction and medication recommendation across MIMIC-III and MIMIC-IV.

Audio-Visual Large Language Models (AV-LLMs) remain prone to cross-modal hallucinations, where one modality incorrectly affects predictions about another. Although contrastive decoding reduces hallucinations in vision-language models, its direct extension to AV-LLMs overlooks a key challenge: different questions require different perceptual evidence, including audio, video, or their interaction. Notably, we observe that joint audio-visual inference can weaken the prediction even when a model can recover the correct answer from a single informative modality. For example, when asked which instrument is heard, a model may correctly predict violin from the audio alone. Once a video showing a guitar is added, its confidence in violin may drop. In this paper, we introduce Relevant Evidence Decoding (RED), a training-free method that identifies question-relevant evidence and selectively strengthens its contribution. RED uses pointwise mutual information to quantify the predictive support provided by audio and video beyond the question alone. It decomposes their joint contribution into audio, video, and residual interaction components. A question-only inference pass determines the required evidence type, after which the model augments the original audio-visual prediction with the corresponding PMI contribution. Across three audio-visual hallucination benchmarks and three AV-LLMs, RED improves accuracy over standard decoding by up to 7% on CMM, 6.3% on AVHBench, and 3.8% on SVHalluc, with an average relative time to first token of 1.5x standard decoding.

Large language model (LLM)-based self-evolving search is a promising approach to scientific discovery. However, high-fidelity evaluation of every candidate is prohibitively expensive in some domains. Self-evolving systems in such settings therefore rely on low-cost but imperfect proxy rewards, which may assign high scores to infeasible candidates. These false positives may contaminate both the final output and the feedback used to guide subsequent generations. This motivates statistically calibrated reward intervals for more reliable self-evolving search. We propose Conformal Interval-Driven Self-Evolution (CISE), which constructs candidate-specific reward intervals using conditional conformal inference and iteration-wise online density-ratio estimation. CISE uses conservative interval-based rewards for evolutionary feedback and returns candidates only when all required property intervals lie entirely within their respective feasible regions. We derive fixed-iteration coverage results under explicit assumptions of independence and covariate shift. We evaluate CISE on three self-evolving search tasks in materials science. In our experiments, all candidates returned by CISE are true positives under high-fidelity evaluation, whereas the baselines return more candidates but include false positives. These results highlight the value of a smaller, more precise shortlist when downstream validation budgets are limited. Our repository is available at https://github.com/MLAI-Yonsei/CISE.git.

Image-based traversability estimation is inherently dependent on the robot platform, deployment domain, and mission preferences, which limits the applicability of purpose-trained models. To facilitate domain adaptation, this work aims to reduce the number of required annotations in the target domain using sample-efficient preference learning. Our method represents traversability through von Mises-Fisher mixture prototypes in a frozen vision-language feature space. Relative natural-language rules provide a commonsense prior, while sparse relative image annotations adapt the prototype directions and utilities to a target domain through computationally and sample-efficient fine-tuning. Experiments on WayFAST demonstrate accuracy competitive with end-to-end trained estimators while enabling sample-efficient image-based adaptation. Qualitative experiments further demonstrate the language prior's zero shot applicability and the fine-tuned estimator's improved dense prediction on semantic maps. Evaluation is complemented via semantic interpretation of learned prototypes by dissecting semantically close natural language prompts. Code and trained estimators available at https://resireg.github.io

We present a variant of the multiset path order for untyped applicative term rewriting. Compared to existing work, our variant incorporates two distinctive features, dubbed arity assignment and reification, to overcome difficulties in handling partial and variable application.

Large language models (LLMs) can support rubric-based screening of CVs, but applying a high-capability model to every candidate and criterion is costly. We present CreateScore, a domain-theory-informed Bayesian network for criterion-level LLM routing. A hand-specified directed acyclic graph with Dirichlet-multinomial conditional probability tables converts CV evidence into posterior uncertainty; low-uncertainty decisions are resolved by a local 8B model and uncertain ones are escalated to a 120B reference model. The graph is causally motivated, but the system performs standard Bayesian conditioning, not causal inference. The escalation threshold is calibrated on a training fold (target: 70% resolved locally) and then frozen. On 200 synthetic Data Science CVs (139 training and 61 test candidates, five criteria), 77.7% of criterion decisions were resolved locally (237 of 305). Relative to a reference condition in which the 120B model adjudicated every criterion, routed escalation reduced token use by 65.2% and raised exact score agreement from 32.8% (8B alone) to 42.6% (95% CI 31.0-55.1%); at n = 61 the gain was not statistically distinguishable. The uncertainty signal did not, however, identify the decisions on which the 8B model erred: disagreement with the reference was 16.2% among escalated and 19.4% among locally resolved decisions (AUROC 0.47, 95% CI 0.39-0.56), no better than random selection. We also document how an earlier evaluation was invalidated when truncated reasoning-model outputs were silently replaced by local labels, and we recommend safeguards for cascade evaluation. CreateScore is supported as an auditable cost-reduction mechanism, not yet as a targeted error detector, and is not an autonomous hiring system.

Understanding human cognitive activity in everyday life remains challenging due to the dynamic, context-dependent, and multimodal nature of cognition. Laboratory-based studies often fail to capture real-world cognitive processes, while single-modality approaches provide only partial insight into cognitive states. Advances in wearable sensing now enable the collection of heterogeneous data streams for a more comprehensive view of daily cognition. This paper presents a multimodal framework for tracking human cognitive activity using electroencephalography (EEG), wearable physiological signals, behavioral context, and self-reported measures. A preliminary pilot study was conducted with observational data from three participants (N = 3) over 280 annotated 10-minute intervals spanning nine activity domains across two weeks. Results reveal consistent temporal patterns, including a discernible mid-day decrease in motivation and energy at 13:00, followed by afternoon recovery. Work and IADLs yielded the highest flow state rates (51% and 50%), while ADLs produced the lowest (15%). Motivation correlated strongly with arousal (r = 0.78) and attention (r = 0.74), whereas perceived stress showed a weaker negative relationship (r = -0.33). A linear regression model predicting motivation from arousal, attention, energy, and stress achieved R2 = 0.76 (MAE = 9.84, RMSE = 12.85). Lag-based analysis indicates that prior energy levels positively predict subsequent motivation, confirming temporal dependencies in cognitive dynamics. These findings demonstrate the feasibility of multimodal cognitive activity analysis in real-world environments and highlight the importance of integrating physiological and behavioral indicators. The proposed framework provides a foundation for future large-scale multimodal systems and applied intelligent solutions.

Large language models (LLMs) have shown promising potential for biomedical named entity recognition (BioNER) through instruction following and in-context learning. However, existing LLM-based BioNER methods still face two key limitations. First, retrieved demonstrations and external biomedical knowledge provide limited support for dataset-specific annotation semantics, leaving entity boundaries, type scopes, and annotation conventions ambiguous. Second, free-form generation lacks sufficient structural control, often leading to invalid formats, hallucinated mentions, duplicated entities, and boundary errors. To address these limitations, we propose GAMA, a guideline-augmented multi-agent framework for schema-as-code BioNER. GAMA first induces candidate annotation rules from labeled training instances and verifies them against annotated data to construct reliable dataset-specific guideline memory. Guided by these verified rules, a planning component generates ranked span-type hypotheses with rationales, and a coding component converts them into schema-constrained entity objects. A verification module then checks span grounding, type validity, and structural compliance, and performs dual-loop refinement to correct invalid or low-confidence predictions. Experiments on three widely used BioNER datasets using two different LLMs demonstrate the effectiveness of GAMA, achieving the best performance in four out of six LLM-dataset combinations. Ablation studies further validate the contributions of its key components.

Recipe generation from food images has practical value for intelligent cooking assistance, but traditional one-pass generation often overlooks user-specific safety requirements such as allergies, dietary restrictions, and preparation constraints. We propose PCAR, a Planner-Critic Agentic Remediation framework for trustworthy recipe generation. PCAR separates recipe planning from safety verification: a Planner Agent extracts ingredients and generates recipe drafts conditioned on the user profile, while a Safety Critic Agent audits each draft and provides structured feedback for remediation when violations are detected. This remediation loop enables unsafe recipes to be revised rather than directly returned or discarded. We evaluate PCAR on real food images with 100 benchmark user profiles across four backbone models, including proprietary and locally served open-source models. Results show that PCAR achieves strong safety and generation performance with capable backbone models, while preserving practical recipe quality.

Large language models (LLMs) can infer user preferences from interaction histories and reviews, yet the rationales they generate may not reflect the information actually used for recommendation. A preference claim may be weakly supported by its selected evidence, or may have little effect on the final ranking. We refer to these two failures as the grounding-influence gap. We introduce PROVE-REC, a general framework for verifiable preference reasoning in LLM-based recommendation. Pass A converts the complete pre-target history into a compact preference proof consisting of positive and avoidance claims linked to selected evidence entries. Pass B predicts the next item using only the proof and its selected evidence, preventing the recommender from bypassing the reasoning path. To verify evidence-to-proof grounding, we compare the effect of masking selected evidence with masking a comparable control entry. To verify proof-to-recommendation influence, we remove a preference claim and measure the resulting decrease in the target item's ranking margin. A ranking-preservation objective further retains useful information from the complete history. Comprehensive experiments on wide-ranging real-world datasets demonstrate that PROVE-REC consistently outperforms strong sequential, generative, and LLM-enhanced baselines, with improvements of up to 7.45%. Controlled ablations confirm the effectiveness of the two-pass architecture and verification objectives. Moreover, PROVE-REC produces claims that are more strongly grounded in historical evidence and more influential to recommendation while preserving ranking quality.

Recent text-to-image generation models have achieved remarkable visual quality, but improving them through post-training remains challenging because no single reward signal captures the full range of human preference. In this work, we develop a simple and effective post-training recipe for open-domain text-to-image generation based on the composition of complementary reward signals. Our reward system consists of two main components: a preference reward, trained on large-scale human preference data using a Bradley-Terry objective to capture overall human aesthetic and perceptual preferences, and rubric-based rewards, which explicitly evaluate prompt faithfulness and other desirable properties while providing safeguards against reward hacking. A key challenge is how to combine these heterogeneous reward signals. We show that a naive weighted average leads to suboptimal optimization behavior, and propose a simple reward composition strategy that more effectively balances preference optimization with rubric satisfaction. In the Arena text-to-image leaderboard (https://arena.ai/), our RL-trained Flux2dev achieves an Elo rating 69 points above the base model, and our post-trained Ideogram-4 surpasses every open-source model on the leaderboard, reaching an Elo of 1223.5. (Claims of state-of-the-art performance are based on the Arena leaderboard snapshot as of September 4, 2026.) Our results suggest that effective rewards for frontier generative-model training require broad coverage of user intent and robustness to exploitation under optimization. To support reproducible research, we release Arena-T2I-Training, a 1K subset of training data that recovers some gains of full-scale training, providing a resource that we hope will facilitate future work on post-training for text-to-image models.

Deep learning has shown remarkable success in the data-driven modeling of dynamical systems. Much of its success is attributed not to the flexibility of neural networks but to inductive biases based on physical prior knowledge, such as energy conservation and symplecticity. However, existing methods do not fully exploit the fact that real-world physical systems are interconnections of components. Some methods require the interconnection to be known a priori, while others assume the system to be reducible to an ordinary differential equation (ODE) and learn only the reduced ODE, discarding the algebraic constraints imposed by the interconnection. Here, we propose Dirac-interconnected neural elements (DINEs), a neural network model that represents a physical system as a differential-algebraic equation (DAE), whose algebraic constraints are given by a Dirac structure in kernel representation. With DINEs, we simultaneously identify from data the interconnection among the components as a Dirac structure and learn the characteristics of the components as neural networks. This allows us to keep the learned subsystems in unreduced form and isolate or compose them to make a new system without retraining. Moreover, DINEs can handle partially observable systems. Experimental results demonstrate these capabilities on physical systems beyond the reach of existing methods.

Recent text-to-image models have made substantial progress in photorealism, aesthetics, and text-image alignment. Yet visually appealing images can still violate real-world plausibility, exhibiting malformed object structures, impossible anatomy, physically implausible interactions, or inconsistent spatial relationships. Such failures are not well captured by existing fidelity, aesthetics, preference, or alignment metrics. To address this gap, we introduce TerraVis, a framework for evaluating world-grounded visual consistency in generated images. TerraVis defines a structured taxonomy of world-consistency violations spanning object-, interaction-, and scene-level failures, and employs a multi-stage evaluation framework to identify and quantify them. Given an image, TerraVis first uses an MLLM to assess its eligibility for evaluation, then detects violations across 18 taxonomy-defined types and classifies them as minor or major to derive an overall world-consistency score. Across diverse open-source and proprietary text-to-image models on two widely used benchmarks, TerraVis achieves the strongest correlation with human judgments of world consistency among existing metrics. Our benchmark results further show that models that achieve strong performance on conventional metrics can still exhibit substantial world-consistency failures. These findings highlight world consistency as a complementary evaluation dimension and demonstrate that TerraVis enables systematic quantification, diagnosis, and comparison of such failures. Our code is publicly available at https://github.com/ShyFoo/TerraVis.

World models learn state evolution from trajectories, making access to temporal context a central training requirement. Federated learning can use distributed records, while ownership boundaries within a trajectory restrict the examples each client can construct. Our study benchmarks this cross-time setting through hourly action-conditioned clinical prediction on eight MIMIC-IV disease cohorts, comprising 40.87 million transition memberships. We specify severity-based client ownership, patient-separated construction, local history and future-window rules, and paired rollout evaluation from one to 32 hours. A matrix of ten federated algorithms covers 32 disease–partition configurations under five rounds of ten-percent participation. Three findings emerge from existing results and training logs. First, client ownership and participation jointly restrict long-window coverage: only 7.55%–21.36% of pooled-available 32-step windows have a locally complete anchor visited during training, averaged across diseases. Second, finer severity partitions accompany higher FedAvg error in 15 of 16 paired comparisons, while algorithm gains are small and horizon-dependent: FedProx reduces mean error by 0.56%, with no consistent improvement at 32 steps. Third, algorithm labels conceal distinct update behavior, including inactive extrapolation and orders-of-magnitude differences in update scale. Cached-update performance also varies strongly across trajectory partitions under the same benchmark protocol. These results establish temporal access, participation coverage, optimization behavior, and horizon-resolved prediction as complementary dimensions for evaluating federated clinical world models.

Reconstructing adversarial behavior from Industrial Control System (ICS) telemetry is difficult because process observations reveal physical changes more directly than the actions that produced them. This paper presents a Digital Twin (DT)-assisted framework that extracts synchronized state changes, converts them into evidence-preserving descriptions, maps them to ATT&CK for ICS through retrieval-augmented Large Language Model (LLM) reasoning, and constructs a typed dependency graph. Evaluation comprises a four-configuration ablation on nine held-out SWaT scenarios containing ten telemetry-evaluable ground-truth episodes, three independent generations of the principal mapping configurations, and external evaluation on BATADAL and WADI. Across the three SWaT generations, DT-enriched mapping produces fewer annotation-relative False Positive (FP) episodes in every run, with a mean per-run reduction of 34.1%. Both configurations obtain a mean recall of 0.633, although recall varies between 0.50 and 0.70 and DT enrichment does not improve F1 in every run. These observations are descriptive: the primary-run paired comparisons do not reach statistical significance, and the mapping advantage does not transfer to either external dataset. An exploratory positive-only dependency benchmark recovers eight of nine documented Co-occurrence relationships with DT context. A separate end-to-end benchmark containing one positive pair and 25 negative controls exposes propagation of mapping errors into unsupported edges. The findings identify both the potential and limitations of DT context for semantic security interpretation, without establishing reliable autonomous attribution, general causal reconstruction, or practical analyst benefit.

Biologically-informed neural networks (BINNs) have emerged as a flexible subclass of physics-informed neural networks (PINNs) for learning terms in partial differential equations from data. BINNs are particularly suited for biological systems, where the governing equations are highly nonlinear and only partially known a priori, and where data observations are often sparse, noisy, and incomplete. However, applying BINNs effectively in practice depends critically on hyperparameter selection, which remains a central challenge in equation-learning frameworks. Hyperparameters are often chosen heuristically and only cursorily documented, which limits the reproducibility of results and the transferability of methods. We present a diagnostic workflow for hyperparameter selection that can be used when the ground-truth equations are not known. The workflow is guided by three main questions: (1) Are the benefits of greater network capacity worth the cost? (2) Do more training epochs keep reducing the validation loss? (3) Do the learned terms stop changing as network capacity and training increase? We apply our workflow to synthetic systems of varying complexity with known ground truth, spanning diffusion and growth right-hand side terms and data ranging from 1D+t to 2D+t. We demonstrate that the validation loss generally follows the true error in the learned terms and distil practical rules of thumb for selecting hyperparameters in the BINN architecture. By providing a structured workflow, practical guidelines, and suggested starting values for hyperparameter selection, this work lowers the barrier to BINN-based equation learning.

Long-context LLM serving is increasingly bottlenecked by KV-cache memory, especially in resource-constrained scenarios. Among existing KV-cache compression strategies, token-wise methods reduce cached states but risk information loss through eviction or condensation, while feature-wise methods reduce per-token KV dimensions but can require full-dimensional reconstruction to apply positional embedding, limiting decoding speedups. We introduce SlimKV, a question-agnostic joint token-feature KV-cache compression method. SlimKV uses low-rank-aware training to compress long contexts into beacon memory states with latent KV representations, together with layer-adaptive rank allocation. We further uncover a positional asymmetry: removing key-side RoPE affects beacon and raw tokens differently, with much smaller degradation for beacon tokens. Exploiting this asymmetry, SlimKV trains beacon KV projections under a K-RoPE-free constraint and enables latent-space attention during decoding, mitigating reconstruction latency. On LongBench, SlimKV outperforms baselines at 16x/32x compression and remains leading at 4x/8x, where it retains over 96% of the uncompressed model's score. Needle-in-a-Haystack confirms robustness across evidence positions, and efficiency evaluation shows up to 7.34x attention speedup and 3.38x end-to-end decoding speedup over the uncompressed model at 128K length.

Test-driven development gives AI coding agents executable requirements for implementing software. Because these agents can adapt their implementations to the examples they observe, passing a predetermined collection of tests can leave substantial parts of the intended behavior unimplemented. We propose Generative Test-Driven Development (GTDD), a formulation of test-driven development in which a separate testing agent generates new inputs after each candidate implementation is fixed, using a human-specified behavioral contract and the feedback from earlier rounds. A trusted evaluator checks these inputs, returns reduced counterexamples to the coding agent, and saves them for regression testing, so development continually confronts failures beyond the initial examples. We characterize the evidence that this process provides through a finite-population analysis of false acceptance under adaptive candidate selection. The resulting bounds quantify how test visibility and repeated feedback affect acceptance, and show that fresh random audits after candidate commitment control false acceptance across development rounds. In a paired experiment on a stateful key-value store, both policies that regenerated tests during development ended with lower mean failure rates than the policy whose tests were generated once by the same language model, and giving the tester the candidate's source produced no detectable additional improvement. Further conditions requesting equal numbers of tests did not isolate any single feature of the policies as the source of this difference. GTDD combines this adaptive development feedback with established regression tests and an independent acceptance rule.

Mixture-of-Experts (MoE) architectures scale language models efficiently by activating only a few experts per token, but the saving is confined to computation: every expert must stay resident on the accelerator, so memory bounds where these models can be deployed. Expert pruning reduces this footprint, yet existing methods are static — a single mask, calibrated offline, is applied to the model for every subsequent request. This assumption can fail when the workload is heterogeneous, most prominently in multi-agent systems, where one backbone serves many tasks and roles at once: our analysis shows that different tasks and roles recruit different experts, while static methods assign one fixed subset to all of them. We therefore propose Dynamic Expert Pruning (DEP), which rests on a finding we establish here: an agent's system and task prompts are by themselves sufficient to identify the experts that agent and its task require, since that text already describes what the agent will do. A lightweight predictor, trained once on workflow transcripts, turns those prompts into a specialized per-request mask in a single forward pass, with no per-configuration calibration. Across diverse tasks and roles, model scales, and MoE architectures, DEP achieves better overall accuracy than static pruning and merging baselines, and generalizes to workflows unseen in training without retraining. Its margin over those baselines is largest when few experts are retained, suggesting that the role specialization inherent to multi-agent systems permits sparser serving than static pruning allows.

Finite-state interval Markov-chain (IMC) abstractions provide sound verification and performance bounds for continuous-state stochastic systems by enclosing cell-to-cell transition probabilities and costs in intervals. Finite-state Markovian abstractions are generally lossy, as state aggregation often destroys the Markov property. Recently, memory-dependent Markov-chain (MC) abstractions of stochastic systems have been developed to mitigate this by recording recently visited cells. In this paper, we develop the memory-dependent extension of IMC abstractions. Building on the insight that memory narrows the family of state distributions that could occur inside the current cell, we prove that memory tightens the local intervals, at the price of a larger abstraction. We also formulate no-memory IMC abstractions, tightening classic IMC abstraction intervals without enlarging the state space. To characterize spatial refinement relative to memory, we derive two cost-guarantee tightness upper bounds. The bounds depend, respectively, on cell size, and on the worst-case remaining distributional ambiguity after filtering over the remembered past. We then specialize our result to linear-Gaussian systems, for which we derive a closed-form upper bound on this ambiguity. Numerical examples show that the memory-dependent and no-memory constructions can produce tighter expected-cost intervals than comparable partition refined models.

Instruction tuning has become a common paradigm for applying large language models (LLMs) to biomedical named entity recognition (BioNER). However, existing instruction-tuning approaches still face two key challenges. First, conventional natural-language instructions typically serialize BioNER annotations as flat textual outputs, providing limited structural constraints for typed entity extraction. Second, high-quality biomedical annotations are limited, and learning from a single serialized output form may restrict structural diversity and reduce model robustness. Although external biomedical knowledge can be introduced to alleviate data scarcity, it often requires costly resource construction. To address these challenges, we propose MITE, a Multiple Programming Languages Instruction Tuning and Ensemble method for BioNER. MITE reformulates BioNER as a structure-to-structure generation task by representing both instructions and entity outputs in code-formatted representations. Specifically, each training instance is transformed into multiple programming-language formats, including Python, C++, and Java, while preserving the same underlying entity semantics. These language-specific representations provide structurally diverse supervision without requiring external biomedical knowledge or additional annotations. During inference, MITE aggregates predictions from different code formats through an entity-level voting strategy, reducing language-specific prediction variance and improving robustness. Experiments on three widely used BioNER datasets demonstrate that MITE achieves competitive performance against representative BERT-based and LLM-based baselines, attaining the best results on two benchmarks. Ablation studies further validate the effectiveness of multi-language representations.