Papers

Accurate indoor localization is essential for emerging applications in robotic navigation and search and rescue. While classical methods typically focus on single-point estimates, complex indoor environments with heavy blockage and multipath propagation often lead to multimodal likelihood surfaces where a single estimate is insufficient. This paper proposes LOCUS-DT (Localization via Observation-Conditioned Uncertainty Scoring with Digital Twins), a framework that treats snapshot localization as posterior inference over the transmitter location. By leveraging a ray-tracing-based digital twin (DT) of the known environment, LOCUS-DT generates synthetic multipath profiles for candidate locations and compares them against the measured channel profile. Central to our approach is a novel learned scoring function designed to compare a fixed number of dominant specular paths, providing robustness against errors in both the DT environment model and the physical channel estimation. Importantly, LOCUS-DT is trained over an ensemble of environments to ensure generalization to unseen layouts. We evaluate the system using a Sionna-based ray-tracing backend, demonstrating that LOCUS-DT captures the sharp, multimodal posterior structures inherent in indoor settings more accurately than standard Gaussian or Gaussian-mixture benchmarks.

Estimated Time of Arrival (ETA) prediction is a core component of intelligent transportation systems. As traffic congestion patterns become increasingly dynamic in large cities, maintaining high prediction accuracy poses a major challenge for ride-hailing platforms. Existing methods either fail to adapt to irregular traffic patterns and sudden congestion, or suffer from new distributions without disentangling long-term trends from short-term fluctuations, thereby degrading model performance in real-world scenarios. To address this challenge, we propose DSETA, an incrementally updated Dual-Stage ETA prediction framework. Specifically, the continual learning process is divided into inter-day and intra-day stages. We first design the intra-day learning stage, which relies entirely on real-time data to enable dynamic adaptation to short-term traffic patterns caused by events like holidays or accidents. Next, we develop the inter-day learning stage, which leverages aggregated historical data from a short time window to capture knowledge of long-term distribution shifts, such as seasonal trends and traffic network evolution. Subsequently, to prevent catastrophic forgetting and preserve knowledge of regular patterns, we explore a Historical Traffic Knowledge Consolidation module. Finally, we validate DSETA's effectiveness and robustness through extensive offline and online experiments conducted on real-world datasets from DiDi's platform. Online A/B tests across three major cities including Beijing, Wuhan, and Xi'an consistently demonstrated performance gains, achieving MAE reductions of 6.62%, 0.73%, and 2.40% respectively. This framework has been successfully deployed in DiDi's production environment, processing hundreds of millions of daily requests and validating its strong performance in industrial applications.

Nonlocal partial differential equations arise in many applications but are often difficult to model and learn because of the presence of nonlocal operators. We present a flow-map learning (FML) framework for modeling unknown nonlocal PDEs directly from solution data. Rather than learning or approximating the underlying nonlocal operators, the proposed approach learns the finite-time evolution operator in either modal or nodal space. Two complementary formulations are developed for spectral and grid-based solution representations. Numerical experiments on one- and two-dimensional fractional diffusion and wave equations demonstrate accurate and stable long-time prediction using only short observation windows. The proposed approach provides an effective data-driven framework for learning unknown nonlocal dynamics without explicit evaluation of nonlocal operators.

Logical rule induction seeks interpretable rules that transfer across propositional schemas. This requires respecting symmetries: atom naming, example order, polarity flips, and label swap. Enforcing exact symmetry by construction lets one trained inducer scale beyond its training schemas. Our central contribution is a canonical export that decodes a discrete rule from literal scores. It needs no retraining and is exactly equivariant whenever those scores respect the symmetries. We instantiate it on the Neural Rule Inducer, a disjunctive-normal-form (DNF) foundation model that natively respects only example order. We restore the remaining symmetries through architecture, inference, and training. On synthetic stress tests, accuracy on the support labels stays stable at much larger schemas, and rule fidelity on fresh inputs remains above the unmodified model. On real data, accuracy improves most on larger schemas. The exported rule is exact on synthetic full-group tests and on schema-valid real-data tests. This is a mathematical property of the export rather than of a specific model, and we validate it empirically only on the NRI. Enforcing symmetry by construction turns this small-data pretrained model into a reusable, interpretable inducer that transfers to larger schemas.

Claims data can show that provider behavior changed but cannot by itself explain why. Fraud, waste, and abuse (FWA) review requires identifying material behavior, locating the codes and dollars driving it, and testing plausible explanations. Forecast residuals conflate growth, service-line shifts, code maintenance, and incomplete observation with potentially concerning behavior. We instead formulate provider review as a descriptive representation problem: observed amount $y_{ijt}=s_{it}p_{ijt}$, where $s_{it}$ is provider scale and $p_{ijt}$ is procedure composition. The profile records scale history, effective-dated code lineage, clinical-family shares, first-use events, billing context, and Medicare-versus-client differences. An optional rank-32 nonnegative factorization of procedure co-occurrence adds a fixed semantic geometry for similarity and retrieval. The profile surfaces evidence for review without inferring intent or adjudicating FWA, and is one engine within Falcon's broader review system. In a ten-quarter proprietary Medicare Carrier and DME build (1.27 million providers; 9.2 million provider-quarter profiles through 2026~Q2), the semantic dictionary covers 3,641 procedures and raises recall at 10 from 36.0% to 44.4%; for high-cost rare events, recall at 50 is 56.1% versus zero for popularity. Under the governed eligibility contract, 414,093 providers enter the national review population, with 81.8% and 90.8% remaining eligible across adjacent quarters. Replication across eight client panels preserves 86.9–95.2% amount-weighted semantic coverage. Transparent descriptions thus form the core, with learned representations adding optional semantic context.

Grover search, amplitude amplification, and quantum counting all rely on the same reusable subroutine, a phase oracle, whose construction the algorithms literature takes as given: the classical predicate is assumed to be already encoded as a correct, resource-bounded circuit. We turn this assumption into a measured capability. QEncodeBench tasks large language models (LLMs) with encoding classical constraint problems as phase oracles and scores the generated circuits with an adversarially self-validated verifier that decides full solution-set equivalence up to a global phase, with ancillas restored and resource budgets enforced. Sampled basis-state tests, we show, systematically overestimate this ability. Measured this way, models separate sharply: code models without a reasoning mode solve essentially nothing, and enabling native reasoning on identical weights improves accuracy by an order of magnitude. The failures are overwhelmingly semantic rather than syntactic. Two architectures, a unit-verified constraint agent and a neuro-symbolic compilation pipeline, close most of the remaining gap by delegating correctness-critical composition to deterministic procedures. Ablations quantify the contribution of each component, and resource gating exposes an architecture-dependent trade-off between circuit width and depth. Finally, controlled difficulty escalation reveals architecture-specific responses to difficulty structure: different difficulty axes degrade different methods, while the neuro-symbolic pipeline passes every evaluated instance. Code and data are available at https://github.com/chexujun/QEncodeBench.

Point-of-Interest (POI) recommendation is a core task in location-based services, yet most existing methods assume a fixed user population and POI catalog. Through a large-scale data-driven analysis of 10 U.S. cities, we identify substantial POI churn, user turnover, category drift, and decay in static POI memory, motivating the study of continual dual cold-start POI recommendation. To address this setting, we propose EviRec, a continual evidence-learning framework that estimates how much historical evidence should be trusted separately for each candidate POI. EviRec scores each visible candidate from three complementary views: a matching view based on the user's recent mobility profile, a transition-memory view that captures repeated mobility routines, and a lifecycle view that reflects candidate maturity. Because a near-zero transition score may indicate either irrelevance or insufficient observation, EviRec qualifies the evidence using each candidate's observation state and applies a reliability gate to adaptively route between transition-memory and lifecycle evidence. We evaluate EviRec on a full-year, five-city POI check-in dataset containing more than 30,000 users and 684,200 trajectories. Experimental results show that EviRec consistently outperforms state-of-the-art baselines, with the largest gains concentrated on cold-start queries. In particular, EviRec improves NDCG@10 by 20.4% on Dual-New cases over the strongest baseline. In-depth analyses further confirm that these gains arise primarily from candidate-specific reliability gating while largely preserving previously learned mobility routines.

Human activity traces record individuals' timestamped visits to points of interest and are essential for applications such as mobility prediction and urban simulation. However, accessing large-scale HATs is challenging due to high collection costs and privacy concerns. Synthetic HAT generation offers a promising way to make such data available and has attracted growing interest from both industry and academia. Although many efforts have been devoted to this topic, most of them rely on real data from a region to generate synthetic data for the same region, which is infeasible for the many regions where real HATs are unavailable. To fill this gap, we propose ZeroHAT, a behavior-conditioned framework that generates synthetic HATs for a target region in a zero-shot manner by transferring behavioral patterns learned from real HATs in source regions and adapting them with publicly available contextual information about the target region. ZeroHAT has three key novel components: (i) a multidimensional consistency-aware intent extractor; (ii) a cross-region behavioral cloning module; and (iii) a behavior-conditioned activity realization module. We evaluate ZeroHAT on a ten-city benchmark, where extensive experiments show that ZeroHAT achieves 4.5-6.4x the normalized downstream utility of the strongest baseline and improves average fidelity by 15.6%-40.8% across target regions.

Spatially heterogeneous partial differential equations (PDEs) exhibit location-dependent dynamics arising from variations in geometry and physical coefficients. Existing neural operators improve localized modeling through multiscale features, attention mechanisms, or domain decomposition, yet their update rules often remain spatially shared. Hypernetwork-based methods adapt parameters across PDE instances but typically generate only one global parameterization per instance. Consequently, shared operators may underfit boundaries and high-gradient regions, with these localized errors accumulating during autoregressive rollout. We propose a spatially adaptive neural operator (SANO), which replaces this spatially shared parameterization with a spatially continuous field of location-dependent operator parameters. SANO uses Fourier-encoded coordinates and a coordinate-conditioned hypernetwork to generate spatial operator-conditioning codes at sampling points. A Hyper-Neural Element (HNE) mechanism interpolates these codes within local subregions, coupling neighboring operators while allowing their update rules to vary across space, and partition-of-unity weights assemble the overlapping local predictions. Experiments on one-, two-, and three-dimensional PDEs and two perforated-domain elliptic benchmarks show that SANO consistently outperforms competitive neural-operator, hypernetwork-based, and physics-informed baselines.

A recogniser trained on one person rarely transfers to the next, and useful performance usually demands a fresh round of labelled calibration from the end user. A systematic review of 1077 studies quantifies where the evidence is thin: amputees appear in about one in six. Here a montage-agnostic cross-user encoder is carried to eleven transradial amputees on a protocol matched to its intact-limb training data. Zero-shot cross-population transfer fails outright: the encoder requires labeled data from the new user before it begins decoding, and it then exceeds the per-user classifier a clinic would fit by 0.190 macro F1 at three repetitions and for every subject in the cohort. Given three labelled repetitions it reaches 0.779 macro-F1 against 0.589 for the per-user pipeline. Training on forty intact subjects produces better transfers to a new amputee than training on ten other amputees, and combining the two produces better transfers than either individually. The prediction pre-registered for this study, which extends the encoder's baseline-strength account with the premise that amputee EMG is less separable, holds true only after a few repetitions become available and after enriching the source pool with additional amputees. At a single repetition, and at every budget under a source matched to the intact-limb comparison, it fails. Thus, it locates the boundary of the proposed account.

A myoelectric interface needs calibration from the user before it will function. Earlier work has treated calibration as a quantity, but has not asked the question of what a device should do with the calibration repetitions once they have been collected. This paper views personalizing the cross-user encoder as a design decision with a cost. Four alternative approaches to using exactly the same labeled repetitions were tested from a single cross-user encoder per held-out subject. Prototypical adaptation, linear probes, scaled fine-tuning and full fine-tuning were tested at every budget up to the maximum each database allows, five repetitions on DB1 and four on DB2 and DB5. Comparing four ways to spend a small calibration budget across 77 subjects, full fine-tuning is the most accurate at every budget, consistently enough that there is no exception among subsets of subjects. The result which impacts how one might make an engineering decision however is that a gradient free prototypical rule recovers 52 to 78 per cent of its benefit with no optimiser and no per-user copy of the weights, which makes personalisation something a worn device can do at donning time. The widespread intuition that a good representation only needs a fresh classifier is incorrect here. How well each method may perform relative to a per-user classifier that would be fitted by a clinic will depend on the specific database.

Sharpness-aware minimization (SAM) doubles the cost of every training step, yet its benefit concentrates where training ends. We study where an expensive training mode should be spent and propose Temperon: a plain-SGD explorer for the first 43% of the epoch budget, then one scheduled hand-off that gives the entire final cosine anneal to a SAM-wrapped Muon refiner. On CIFAR-10/100, SVHN and Tiny ImageNet (five seeds, times reported as epochs-to-target times an idle-GPU-calibrated epoch cost), Temperon matches the best full-time-SAM recipe on accuracy everywhere while reaching the hardest common target 35%, 34% and 32% sooner on three of the four, and sits a tier above the published SAM+SGD recipe at level cost. Ablations make the attribution exact: the Muon refiner is worth +0.85pp with everything else fixed; the explorer's shape and its restarts are worth nothing, and we withdraw them as contributions. Re-running the closest rival, late-phase SAM, at matched budget shows the frontier: it is fastest to every mid-level target, but the tier the Muon refiner buys (0.83 on CIFAR-100, 0.97 on CIFAR-10) is reached by no SGD-refined method in any seed, and on Tiny ImageNet, where Muon buys no tier, the rival simply wins – the measured boundary of the method. The allocation law transfers to GPT-2 pretraining (full-SAM quality at -29% wall-clock) and GLUE fine-tuning (never worse than full-time SAM at a third of its SAM cost). Two constants organize the economics: skipping SAM early buys a fixed credit, and a Muon epoch costs 1.50x a SAM+SGD epoch on all four datasets. Finally, the hand-off cannot be timed from the trajectory: under cosine schedules the accuracy curve is plateau-then-surge, so the information lives in the schedule, making the scheduled switch principled rather than convenient. Code and a pip-installable implementation are released.

Z-loss has been widely applied to the logits of language-model output heads and sparse mixture-of-experts routers. Z-loss constrains the softmax log-normalizers of these output heads and routers, thereby limiting large-logit excursions, reducing finite-precision roundoff exposure, and avoiding training-loss divergence. These use cases arise in modern Transformer settings where large-vocabulary softmax heads, top-$k$ routing, fused losses, and mixed-precision optimizers interact. Z-loss has typically been understood only as a scalar penalty on the log-normalizer. This paper instead analyzes Z-loss from a backward-pass perspective, focusing on the gradients produced by the Z-loss penalty. The logit-space gradient, which we call the backward source, is injected at the logit boundary of the Z-loss branch of backpropagation; consequently, the backward source's effect depends on the architecture and implementation through which the gradient is transported. We develop a backward-transport view for Z-loss that separates the source's scalar amplitude and softmax shape from the transport factors. These factors include common-shift coordinates, tied-embedding pathways, output-to-hidden gain, fused-loss source consistency, optimizer-facing updates, and top-$k$ router reduction scale. These diagnostics show that nearly identical forward Z-loss values can coexist with distinct logit-space Z-loss gradients and, after architectural and optimizer transport, distinct parameter updates. The transport diagnostics also explain why raw-logit Z-loss can reduce scalar tails without changing output-to-hidden gain and why active-route reductions alter the effective router coefficient. Across evaluations of models in the GPT-2 and Pythia families on WikiText-103 and FineWeb-Edu, architecture-aware variants reduce backward-geometry tails while maintaining comparable validation perplexity in low-coefficient regimes.

Large language models (LLMs) prompt a recurring question in research computing: should classical methods like Naive Bayes (NB) be retired? We benchmark Complement Naive Bayes against zero-shot and few-shot LLMs spanning four model families and a 37x range in scale (27B to a 1T-parameter mixture-of-experts) across text classification tasks. LLMs dominate only in zero-data regimes (98.0% vs 88.2% on Amazon Polarity sentiment), and even that win is contamination-prone: on a low-contamination sentiment task NB beats the zero-shot LLM (81.7% vs 73.0%). However, once labeled data is available (e.g., AG News), NB reaches 89.1% accuracy, statistically indistinguishable from the zero-shot 27B LLM (89.0%) and better than the 397B frontier model (84.8%), at thousands of samples/sec on a commodity CPU. Fine-tuned DistilBERT reaches 90.6% but at far lower throughput than NB at batch size 1 (Table 2). Our measured GPU throughput analysis shows small-LLM batched inference is 40-486x slower than NB CPU inference (the multiplier depends strongly on the host CPU), exposing a structural gap bounded by memory bandwidth, with roughly two orders of magnitude lower energy per sample. For resource-constrained HPC practitioners performing text classification with labeled data, NB remains the optimal choice. We show the decision line is task-dependent (NB reaches LLM parity around $N \sim 10^4$ labels for topic classification, while zero-data sentiment favors the LLM at all N tested) and provide a Kubernetes Helm operator that automates model selection using configurable thresholds and verifiable Prometheus metrics.

As deep learning models are increasingly deployed on resource constrained devices, the demand for efficient model optimization techniques continues to grow. Effective deployment of AI models on edge and low power platforms requires optimization methods that reduce model size and computational cost while maintaining high accuracy. This paper presents Quantization Analysis Tool, a practical system designed to streamline quantization workflows and support performance efficient model deployment. Built on the ONNX framework for broad interoperability, the tool provides detailed layer-wise sensitivity analysis, visualization of weight and activation distributions, and insights to guide precision selection. By identifying layers that are resilient or sensitive to reduced precision, the tool enables developers to make informed trade-offs between model size, latency, and accuracy. Experimental evaluations across multiple neural network architectures demonstrate that the tool effectively improves the quantized accuracy, leading to improved efficiency in real-world deployment scenarios. The tool also provides developers valuable insights into the effects on quantization on the model and its accuracy. This work highlights the tools capabilities, practical applications, and its role in enabling efficient AI model deployment through robust quantization analysis

Designing high-performance tactical wireless networks under realistic operational constraints gives rise to challenging combinatorial optimization problems, where the evaluation of candidate solutions relies on detailed physical and traffic-aware models. Although classical metaheuristics such as Tabu Search offer effective mechanisms for exploring large search spaces, their computational cost remains high because numerous candidate moves must be evaluated at every iteration. In this paper, we propose a data-driven framework that improves the efficiency of Tabu Search by learning to guide its move selection process. Rather than altering the neighborhood structure, our approach exploits the information contained in the search trajectories generated during the optimization process. At each iteration, we record both improving and non-improving edge-based transformations together with a set of descriptive features capturing the structural, geometric, and performance characteristics of the network. This information is used to train a Graph Neural Network (GNN) that predicts the impact of candidate moves on the objective function. The trained model is then integrated into the Tabu Search algorithm to rank candidate transformations according to their predicted quality, thereby reducing the number of costly objective evaluations while maintaining an effective exploration of the search space. Experimental results on synthetic benchmark instances demonstrate that the proposed learning-assisted Tabu Search notably reduces computation time while consistently producing higher-quality solutions than the standard algorithm. These findings highlight the potential of combining machine learning with metaheuristics by leveraging the implicit knowledge embedded in search trajectories, paving the way for more efficient solution methods for large-scale network design problems.

Time-lapse electrical resistivity tomography (TL-ERT) provides spatially distributed information on subsurface hydrologic changes. However, inversion of long monitoring sequences is computationally demanding. Modifying the data misfit, regularization, model parameterization, or petrophysical transformation may also require new gradient derivations and separate implementations. Here, we present AD-TLERT, a unified, GPU-accelerated framework for time-lapse ERT inversion based on automatic differentiation. The framework integrates model parameterization, differentiable petrophysical transformations, forward modeling, data misfit, regularization and auxiliary constraints into a single computational chain. Alternative inversion formulations can therefore reuse the same PDE derivative implementation without re-deriving the complete ERT sensitivity for each case. Comparisons with pyGIMLi showed close agreement in the forward responses, gradients, and recovered resistivity models. Under the tested configuration, AD-TLERT achieved an approximately 51-fold speedup. Synthetic experiments showed that inversion choices affect the amplitude, geometry, and temporal behavior of recovered anomalies. By propagating gradients through the embedded petrophysical relationship, AD-TLERT enabled direct water-content inversion and yielded more accurate estimates than post-inversion conversion for the tested model. A field application further demonstrated how ERT, temperature, and soil-moisture observations can be combined to image snowmelt-driven hillslope wetting. AD-TLERT provides an efficient and flexible framework for time-lapse ERT inversion and hydrologic interpretation.

This study proposes a ring-based SpatialTransformer to learn how building uses at different distances from a railway station interact to generate pedestrian flow. Concentric ring buffers at 100-meter intervals up to 800 meters were defined around 100 randomly selected stations in Tokyo, treating each ring as a spatial token. Self-Attention was applied to learn inter-zone interactions directly from data, without prior structural assumptions. GPS-derived walking trip counts served as the target variable and Geographically Weighted Regression as the baseline. Across 30 independent trials, the SpatialTransformer consistently outperformed GWR in predictive accuracy. SHAP analysis revealed that mid-to-outer distance zone features dominate pedestrian flow prediction, while features from the 0-100m zone contributed little. The attention matrix showed that each distance zone attends most strongly to spatially distant zones, demonstrating that pedestrian flow is regulated by structural interactions across the entire catchment area rather than by any single zone in isolation. These findings challenge the compact city assumption that station-proximate development maximizes pedestrian flow, and suggest that land use distribution across the full walkable catchment area deserves greater consideration in urban planning practice.

Large language models for code generation often produce incorrect solutions without reliable indicators of failure. We study whether uncertainty estimation methods developed for natural language transfer to code generation, and whether such signals can improve code generation via selective self-correction. We evaluate five uncertainty methods: mean token entropy, verbalized confidence, $P(\text{True})$, entropy ensembles, and semantic entropy probes, across three small code LLMs on HumanEval and BigCodeBench. We find that multi-sample $P(\text{True})$ achieves the strongest correlation with correctness, while all the other methods, including semantic entropy probes, yield only weak correlation. We then use these uncertainty signals to drive three self-correction policies: adaptive decoding, uncertainty-based regeneration, and verification-based regeneration. Our results reveal a stronger negative finding than anticipated: uncertainty-based self-correction fails to reliably improve Pass@1, degrading accuracy in 5 of 6 configurations across both benchmarks ($-3$pp to $-10$pp), and adaptive decoding degrades accuracy in 4 of 6 configurations. Only verification-based self-correction reliably improves Pass@1, with gains of $+6$ to $+26$ percentage points on HumanEval and $+8$ to $+20$ percentage points on BigCodeBench, scaling inversely with baseline strength. These findings replicate consistently across both benchmarks and suggest that cheap uncertainty estimators are insufficient on their own to improve code correctness, and that their practical value lies in serving as gating signals for costlier execution-based correction loops rather than as standalone substitutes for verification.

We introduce pico-type, a byte-level multi-head content classifier with approximately 1.5 million parameters that simultaneously predicts seven content properties from raw UTF-8 bytes in a single forward pass. Operating directly at the byte level – no tokenizer, no subword vocabulary, no pretrained embeddings – pico-type classifies coarse type (12 classes), modality (8), subtype (24), code language (62), text language (30), file MIME type (90), and risk flags (6-label multi-label: API keys, JWTs, passwords, emails, phone numbers, SSH keys). The architecture combines a learned byte embedding, three convolutional blocks with growing receptive fields, two bidirectional attention layers with rotary position encodings, and a statistical pooling layer feeding seven Matryoshka-style classification heads. Four tiered variants (tiny/small/base/pro) share the same trunk with sliced representations from 16 to 576 dimensions, yielding ONNX exports under 210 KB and CPU inference under 10 ms. Trained on a mixture of synthetic templates and real-world data (8709 GitHub code samples, 5000 Wikipedia articles), pico-type achieves 60.3 percent code language accuracy on The Heap benchmark (24 languages) and 98.2 percent text language accuracy on Wikipedia (30 languages) – improvements of +57 and +79 percentage points respectively over the synthetic-only baseline. Format-based heads (coarse, modality, subtype, file_mime, risk) maintain 100 percent accuracy on synthetic benchmarks. The model, code, and pretrained weights are released under Apache 2.0.

Early identification of lung cancer risk is critical for timely intervention, yet existing prediction models are limited by their reliance on single data modalities and their inability to leverage structured clinical knowledge. We propose LUNG-KGMM, a knowledge-guided multimodal framework that integrates longitudinal electronic health records, radiology reports, chest radiograph representations, and guideline-derived knowledge for 1-to-6-year incident lung cancer prediction. To address modality heterogeneity and potential data leakage, we develop a leakage-sanitized report processing pipeline and a horizon-masked cumulative training objective that handles incomplete follow-up. We further introduce a knowledge-graph representation of clinical guidance that encodes report-triggered finding-attribute-action relations as an auditable knowledge stream. We build a multimodal development cohort from the publicly available MIMIC databases and construct a real-world validation cohort from the Xiamen Medical Big Data Platform. Extensive experiments on the MIMIC cohort demonstrate that LUNG-KGMM achieves superior performance over state-of-the-art methods, and validation on the Xiamen cohort further characterizes its cross-cohort portability and the need for local adaptation. The MIMIC development cohort is publicly accessible; the Xiamen cohort is governed by local data privacy regulations.

Photoplethysmography (PPG) is widely used in consumer wearables because of its low cost and ease of acquisition. However, unlike electrocardiography (ECG), PPG measures peripheral pulse dynamics rather than cardiac electrical activity, limiting its ability to predict cardiac conditions that rely on ECG-specific morphological cues. Existing methods attempt to bridge this gap by reconstructing ECG signals from PPG signals. However, this inverse mapping is inherently ill-posed, and faithful waveform reconstruction does not necessarily translate into improved downstream performance. To address this challenge, we propose P2E-VQ, a retrieval-augmented framework that replaces ECG waveform reconstruction with ECG-linked representation retrieval. Specifically, P2E-VQ converts PPG patches into discrete tokens and retrieves ECG-linked information from a memory bank constructed exclusively from the training data. This process augments PPG representations while requiring only PPG signals during inference. Extensive experiments on five public datasets covering six downstream tasks, including clinical endpoint prediction and affective state recognition, demonstrate that P2E-VQ consistently outperforms pretrained baselines under a unified frozen-feature linear-probing protocol.

Omni-Large Language Models (Omni-LLMs) power complex multi-modal reasoning in applications like World Action Models and autonomous agents. However, their strong performance often masks a profound Perceptual-Decision Misalignment (PDM), where decisions remain unfaithful to multi-modal perceptions. To diagnose this, we formalize Causal Modality Sensitivity (CMS), operationalized via a dual-lens framework: Answer Retention Rate (ARR) at the macro behavioral level, and Logit Angular Discrepancy (LAD) to track microscopic distribution shifts. We also curate CausalMSBench, a diagnostic dataset isolating language priors. Benchmarking reveals that popular Omni-LLMs exhibit critically low CMS, showing negligible distribution shifts even when key modalities are removed. To rectify this, we propose Modality Subspace Activation (MSA), a training-free inference-time framework that uses Singular Value Decomposition (SVD) to estimate modal activation strengths. MSA dynamically balances modal projections in the last hidden state, effectively restoring CMS across benchmarks.

Federated Learning (FL) is a collaborative paradigm that enables multiple devices to train a global model while preserving local data privacy. A major challenge in FL is the non-Independent and Identically Distributed (non-IID) nature of data across devices, which hinders training efficiency and slows convergence. To tackle this, we propose Federated Impurity Weighting (FedImp), a novel algorithm that quantifies each device contribution based on the informational content of its local data. These contributions are normalized to compute distinct aggregation weights for the global model update. Extensive experiments on EMNIST and CIFAR-10 datasets show that FedImp significantly improves convergence speed, reducing communication rounds by up to 64.4%, 27.8%, and 66.7% on EMNIST, and 44.2%, 44%, and 25.6% on CIFAR-10 compared to FedAvg, FedProx, and FedAdp, respectively. Under highly imbalanced data distributions, FedImp outperforms all baselines and achieves the highest accuracy. Overall, FedImp offers an effective solution to enhance FL efficiency in non-IID settings.

Prediction uncertainty is a widely adopted metric for quantifying model confidence, with downstream applications spanning model explanation, data selection, and prediction rollback. Despite its demonstrated utility, the potential of uncertainty quantification to enhance code generation in large language models (LLMs) remains largely underexplored, raising a critical question: to what extent can uncertainty serve as an effective signal for improving LLM-based code generation? To answer this question, we study uncertainty-aware rollback decoding, an inference-time strategy that uses uncertainty signals to identify unreliable generation regions and roll back to earlier valid prefixes without retraining the model. We evaluate this framework on seven code LLMs, five code generation benchmarks, and eight token-level uncertainty signals under a unified decoding setup. Our results show that the complete rollback framework improves over equal-budget restart across the evaluated benchmarks and model settings, with gains of up to 0.26 in pass@1 and 0.35 in AvgTestPassRate on functional code generation benchmarks, and an absolute improvement of up to 6.4% in Patch-Aligned Safe Rate on Dsec-Python. Among the evaluated signals, information-theoretic measures such as token entropy and negative log-likelihood show the most favorable overall trend, frequently achieving the best or near-best results on standard benchmarks. A component-controlled ablation further shows that feedback-guided rollback provides the main improvement, while uncertainty localization provides an additional gain when checking, budget, rollback, and branch decay are held fixed.

The development of 0.1$^{\circ}$ global weather forecasting models based on machine learning (ML) is constrained by the limited availability of high-resolution data, as decades of reanalysis are only available at 0.25$^{\circ}$ resolution. While existing approaches fine-tune 0.25$^{\circ}$ forecast models on limited 0.1$^{\circ}$ samples, we show that this transfer is hindered by the irreversible information loss inherent in coarse-resolution forecasting. Therefore, we propose BaguanHR, a framework that shifts the focus from transferring models to transferring data. We first show that super-resolution (SR) has lower conditional entropy and input amplification than forecasting, making it a more robust vehicle for resolution transfer. By leveraging this advantage through variable-wise SR, we synthesize extensive 0.1$^{\circ}$ data from ERA5. BaguanHR's performance on the synthetic-plus-real dataset exceeds both ML-based methods and IFS-HRES, achieving superior performance across over 85% of the lead times within 72 hours. Furthermore, our findings highlight a power-law scaling effect, as a twofold increase in data reduces RMSE by 4.6% for 72-hour forecasting and 4.9% for 120-hour forecasting. Our results demonstrate that scaling high resolution ML-based forecasting is primarily a data bottleneck, and that variable-wise super-resolution provides a simple yet general solution to unlock long coarse-resolution reanalyses for high-resolution training.

Heterogeneous model fusion seeks to combine models that differ in tasks, initializations, architectures, or scales. We study an underexplored cross-scale setting: improving a small recipient language model with a stronger donor despite substantial architectural mismatch. We ask whether useful capabilities can be transferred without explicit neuron-wise semantic alignment. Building on the observation that truncating a large model to a smaller architecture and injecting it with a tiny mixing weight can already improve the recipient, we propose Activation-Prune-Merge (APM), an activation-guided framework for cross-scale fusion. APM constructs task-conditioned activation maps on the donor, selects salient layers, hidden dimensions, attention heads, and MLP neurons to prune it to the recipient architecture, and injects the resulting donor slice into the original recipient using a micro interpolation coefficient. This formulation treats the donor as a source of concentrated functional components rather than requiring precise structural transplantation. Across 16 benchmarks spanning reasoning, mathematics, code generation, instruction following, and classification, APM improves the overall average accuracy from 55.5% to 60.6% over the original 3B recipient. RTE accuracy increases from 64.3% to 82.3%, QNLI from 52.3% to 65.7%, and BoolQ from 70.8% to 79.2%. Analyses of injection ratios and sequential multi-stage fusion further suggest that activation-guided extraction improves the quality of the transferable donor slice while preserving the small-ratio fusion regime. These results provide evidence that cross-scale heterogeneous fusion can succeed without explicit semantic alignment when the donor contribution is sufficiently concentrated and carefully selected.

A dense, pretrained language model can be retrofitted with recurrent depth and learn an iterative latent transition that persists after outcome-only annealing. Qwen2.5-0.5B-Instruct is split into a Prelude, a weight-tied Recurrent Block, and a Coda, with an identity-preserving one-loop path and a re-entry bridge on later loops. At loop 1 the retrofit remains non-inferior to its base on a preregistered ARC battery. Three findings. First, the mechanism is a reusable procedure rather than terminal-answer lookup, and installs at two budgets: 6M trained parameters over frozen base weights and 180M full-block. With intermediate-step supervision, the model computes one task step per loop and persists when only final answers are graded. The adapter matched the full block overall (83.8% versus 84.0%), led through depth 11, and trailed beyond. Verbal fine-tuning reached 79-86% on controlled verbal renderings (zero-shot transfer was minimal), and adapter verbal training begun from the installed mechanism outpaced matched fresh training by 18.6 points, including on a held-out test set. Second, the operation extrapolates to roughly 1.5 times its supervised depth, holding 70% accuracy through depth 18. Third, a same-size scratchpad-trained model matched the recurrent model within its learned horizon but collapsed beyond it. The recurrent model won overall, 84% versus 72%, retained 53% versus 2.5% beyond depth 10, and answered 7.6 times faster. An iterative transformer can therefore perform deeper reasoning in latent space faster than comparable or larger models fine-tuned on the same task, in a system-level comparison. A second task, running the rule in reverse, exposed the limits: the inverse was learnable in isolation, but no continuation acquired it while preserving the installed mechanism and general capability, a catastrophic-interference boundary. Learned depth selection remains open.

AI weather models are often reported to underestimate extremes, but most evidence concerns deterministic regression models verified against reanalysis. We evaluate twelve physical and AI forecast models against ECMWF IFS using ten months of European station observations. The evaluation covers 10 m wind, 2 m temperature, solar radiation, and precipitation within regimes defined from a fixed ERA5 1991-2020 climatology. We find no uniform AI-specific deficit in the tails. Several AI models remain more accurate than IFS under extreme conditions, while others deteriorate markedly; comparable variation occurs among physical models. Every model nevertheless exhibits a common conditional-error pattern, overpredicting low observations and underpredicting high observations. Attenuation of extreme values therefore does not imply a uniform loss of relative skill: tail performance depends on the model, variable, and evaluation setting rather than on whether the forecast is produced by AI or physical numerical modelling.

An LLM agent's capability depends not only on model weights but on its harness: prompts, tools, skills, and control flow. Self-improvement loops already edit harnesses, yet single-lineage search is path-dependent and local wins often regress other tasks. We introduce DarwinX, which treats self-evolution as selection over a population of harnesses with the model frozen: a preserve-and-extend contract admits only variants that extend coverage without regressing, an archive keeps alternative lineages for recombination, and failure-, teacher-, and self-derived evidence share one edit interface. Fitness comes from each benchmark's own verifier: no gold solutions, no hand-picked winners. Across four benchmarks that progressively separate the evolution signal from the test, one loop adds about 17 points on average: Terminal-Bench 2.1 rises +7.7 to 83.2% on a matched base and to the verified frontier at 84.7% on a stronger one; TerminalWorld's held-out split reaches 68.3%, ahead of every off-the-shelf agent; WebArena-Infinity real-task pass@1 rises from 43.5% to 93.0% audit-clean; and a Terminal-Bench 2.1 harness transfers unchanged to SWE-bench Verified. What evolves is general agent competence, not benchmark-specific patches, so it survives changes of task, verifier, and base model. A frozen model need not be a fixed agent: harness selection turns evaluation compute into durable capability.