Papers

Decentralized learning systems aim to collaboratively train models across multiple clients without relying on a central coordinator. While decentralization improves scalability, privacy, and robustness, it also exacerbates three fundamental challenges: statistical heterogeneity across clients, fairness in client-level performance, and stringent communication constraints. This raises a natural question: how fair can decentralized learning be under limited communication? We address this question by presenting a unified framework for decentralized learning under communication constraints, bringing together graph-based personalization, agnostic fairness, and compressed event-triggered communication. Specifically, we propose a new algorithm DMFL-SQ, a decentralized multi-task learning algorithm that couples personalized model training over a communication graph with an agnostic mixture fairness objective, while reducing communication through sparsification, quantization, and event-triggered synchronization. We establish convergence guarantees for general non-convex objectives and show that DMFL-SQ achieves an $\mathcal{O}(T^{-1/2})$ rate in expected squared Moreau-envelope stationarity despite sparse, quantized, and event-triggered communication. We further derive PAC-Bayes generalization guarantees for the fairness-aware mixture objective. Experiments on CIFAR-10 and the real heterogeneous MUSMET EEG dataset demonstrate that DMFL-SQ substantially reduces communication while maintaining predictive performance and improving fairness across clients. Together, our theoretical and empirical results show that personalization, fairness, and communication efficiency can be jointly achieved in decentralized learning while preserving the dominant convergence rate.

Point-cloud data routinely captured by modern imaging and sensor technologies provide detailed geometric descriptions of objects and environments, but their analysis is hindered by large data volumes, localization noise, and missing information. In addition, existing point-cloud reconstruction pipelines typically return a single best-fit structure without uncertainty quantification. We introduce a fully Bayesian framework for representing point-cloud data and reconstructing closed curves, in which observed points are modeled as noisy perturbations of latent locations constrained to lie on the underlying curve that is regularized by a non-parametric prior. Posterior inference in our framework is carried out using a series of Markov chain Monte Carlo samplers tailored to point-cloud characteristics. Numerical experiments, including synthetic examples and real-world LiDAR datasets, show accurate reconstructions and quantified uncertainty over the recovered curves.

Detecting woody clearing is vital for managing biodiversity. Deep learning models can detect change in woody vegetation from bitemporal remote sensing imagery, however generated products may not meet end-user specifications due to unaligned loss definitions. Further limitations of deep learning models are the reliance on large datasets which can be difficult to attain for spatially rare and ambiguous events such as regrowth detection. In this work we train a model to detect woody change using bitemporal Sentinel-2 imagery consisting of 7 years' worth of annual imagery across the state of New South Wales, Australia. To align the objective of the model with end-user metrics, we introduce the loss scaling coefficient $\alpha$ which transforms the objective to optimize for specific $F_{\beta}$ scores. Introducing $\alpha$ was found to increase precision by 1.85x or recall by 1.12x. We propose input imagery augmentation and generation techniques that allow the woody change detection model to zero-shot transfer to regrowth and woody segmentation tasks. For woody segmentation, image generation techniques using activation maximization with low $\alpha$ values for stability and image generation techniques derived from handcrafted features utilizing a mosaic of clearing patches and artificial trees for contextual grounding were found to outperform prior woody segmentation works of the study area, reducing the overall error by up to 18.2%. For zero-shot woody regrowth, creating pseudo-post and prior images resulted in the model achieving an F1 score of 0.845, creating a foundation for future regrowth detection work.

Updating a graph by inserting or replacing nodes while preserving semantics and reusing existing structure is a recurring computational problem. In population genetics, this problem arises in the genotype representation graph (GRG), a directed acyclic graph that losslessly encodes phased genetic variation across hundreds of thousands of samples by sharing subgraph structure for individual mutations. In a GRG, each mutation's carrier set is implicitly encoded as the set of leaf nodes reachable from the node it is assigned to. Updating a mutation is therefore a structural editing problem, and current approaches remap mutations individually. This paper introduces a batched mutation-remapping algorithm that replaces independent reuse-aware traversals with a single shared reverse-topological pass, identifying reuse candidates for an entire batch at once. The pass propagates compact bit-parallel per-mutation state and uses an adaptive sparse/dense carrier set representation spanning rare-to-common variant densities. Batching is the memory-scalable complement to split-based parallelism, which instead replicates graph and traversal state per worker. Our remapping is evaluated on a controlled update workload and on end-to-end allele polarization, a bulk carrier set update that is common in population genetic analysis. Our approach is up to 10.5$\times$ faster than independent remapping while preserving exact carrier-set semantics.

The resource-task network (RTN) model has been widely applied to represent the technical constraints of complex industrial processes (IPs) such as steel-making, providing the basis for industrial demand response. However, the legacy RTN model contains numerous binary variables and applies different formulations for non-flexible and flexible processes, restricting its computational efficiency and applicability. To systematically improve the computational performance of IP models, we propose continuous RTN model (cRTN), a novel modeling approach that uses continuous variables to represent production tasks and progresses, which are then integrated into unified as well as computationally favorable formulations for the technical constraints in discrete IPs, including resource balance, task execution, waiting time limits, and production targets. Compared to the legacy models, cRTN features fewer binary variables, shorter solving time, and better scalability while maintaining the same accuracy. Numerical tests based on a steel plant demonstrate that cRTN is in typical cases 10 times faster than legacy models and remains tractable with increasing batch sizes, which in legacy models leads to larger problem scales and infeasible solving time. cRTN also achieves a reduction in energy costs by resolving the issue of rounding errors reported in legacy models.

The virtual power plant (VPP) can aggregate flexible resources on the demand side to provide frequency regulation for the grid, helping address the supply-demand balance challenges. When deploying regulation, the VPP disaggregates the requested power adjustment in real time among its internal heterogeneous resources. Achieving optimal power disaggregation in this process is challenging due to the temporal coupling characteristics of the resources, the uncertain regulation signals, and the requirement for fast response. Therefore, existing research relies on heuristic methods, such as proportional disaggregation, and fails to leverage the heterogeneity of multiple resources. Here, we propose an optimal operation strategy for VPPs to provide regulation, exploiting the complementary characteristics of heterogeneous resources by prioritizing the use of low-cost resources while considering temporal coupling. To reduce the computational overhead of online deployment, we further propose a fast disaggregation algorithm to eliminate the reliance on optimisation solvers. We conducted case studies on the operation of a VPP composed of resources including thermostatically controlled loads and industrial production processes. The results verified the reduced operation cost and increased profit of the VPP under the proposed strategy, with only milliseconds of online computation time. We believe that our work can help better exploit demand-side flexibility.

To cost-effectively manage the supply-demand balance of the power system, the flexibility of industrial users could be harnessed through demand-side response. To minimize the negative impact on the production of industrial users during demand-side response, general-purpose models such as the state-task network (STN) are widely used to model the energy-consuming constraints of industrial production processes. However, the required model parameters cannot be set because the required data are privately owned by industrial users and are not directly available, hindering the accurate modeling of industrial loads. In this paper, we propose production scheduling identification (PSI), an inverse-optimization-based approach for industrial load modeling under incomplete information. In PSI, industrial users' smart meter data are used to identify production scheduling parameters, thus addressing the problem of accurate load modeling when private data are unavailable. We implemented PSI with a modified STN and proposed a practical algorithm to obtain an effective solution. Numerical tests showed that PSI can identify the model parameters of a steel powder plant and a cement plant with acceptable accuracy, using only 21 days of hourly smart meter data. Compared with accurate models established with direct access to private data, the modeling error does not exceed 8.5% and 5.2%, respectively.

We study decoding procedures for a family of MDS array codes previously constructed from the Kronecker product of a superregular matrix and a non-singular matrix over a finite field. By exploiting the particular structure of their parity-check matrices, we develop decoding algorithms for different channel models. For the erasure channel, we provide an algorithm capable of recovering any pattern of up to $n-k$ symbol erasures. For the $q$-ary symmetric channel, we investigate the decoding of one and two symbol errors and give explicit procedures for determining their locations and values. We also consider the particular case in which the superregular matrix is a Vandermonde matrix, showing how its additional algebraic structure can be exploited in the decoding process. Explicit examples over different finite fields are provided to illustrate the proposed procedures.

When a single policy is trained in parallel across multiple environments of the same task, such as procedurally generated levels, randomized dynamics, or curricula, implementations commonly use one critic across all sampled environments. Yet different environments can assign different expected returns to the same input visible to the critic. A critic without environment information must then reconcile distinct value targets, systematically shifting the sampled advantages within individual environments. Using illustrative bandit models with multiple environments and a common optimal arm, we characterize how this value mismatch redistributes sampled policy updates, reinforcing unhelpful actions while attenuating or even reversing useful ones. The oracle processes using no baseline, the shared value, or the value specific to the sampled environment have the same mean logit update at a fixed policy and converge to the same optimal policy, yet their realized learning paths can differ sharply. The analysis motivates a minimal intervention: give only a logged environment index to the critic so that it can separate the value targets. Controlled CartPole and MuJoCo experiments expose the predicted shifted values, advantages, and performance gaps. In the more complex BipedalWalker and Procgen settings, the same intervention yields more stable learning and higher returns. Across all $16$ Procgen games, the multihead conditional critic improves aggregate normalized return on $600$ unseen levels per game by $40.8\%$. In conclusion, the theory identifies value mismatch as a direct mechanism through which critic sharing can degrade stochastic learning dynamics, not captured by scalar estimator variance alone, and the experiments show that conditioning on an index is broadly effective in parallel reinforcement learning.

Frontier coding performance is typically attained with large, costly proprietary models. We introduce ledger-based zero-shot self-orchestration (GVS5H), a training-free method in which fresh instances of one model decompose problems and coordinate through a shared file system. Across eleven open and closed-weight models on the 100 latest hard LiveCodeBench problems, the method yields as much as 25.6 points improvement, boosting several cheaper models to frontier-level performance. Orchestrated Qwen3.8 Flash Next scores 93.0% against Fable 5's 90.4% at 9% of the cost, while the smaller Qwen3.8-27B reaches 92.4%. Gains are not universal: some models are unchanged or worse. Transcript analysis attributes the gain to decomposition and persistent context. Inference-time organization can reach or exceed frontier coding accuracy at a fraction of the cost on self-hostable weights.

The problem of finding a fair consensus ranking is an active research topic in the domain of fair rank aggregation and has been well studied; however, existing studies predominantly consider both the input rankings and the output ranking to be permutations where elements are always strictly ordered. In practice, however, a ranking with ties is far more common. This study bridges this gap by presenting a post-processing approach to determine the closest fair consensus ranking when an unfair tie-aware consensus ranking is provided. It proposes an exact algorithm and a fast heuristic to achieve this.

Zero-shot image restoration methods with text-to-image latent diffusion models have achieved great success in universal image restoration tasks without training. However, applying them to video restoration will result in severe temporal flickering. In this paper, we propose a novel framework for zero-shot video restoration and enhancement which uses a text-to-image latent diffusion model and multi-modal references. Through the proposed dual prompt tuning inversion and sampling, the inference time can be reduced to nearly 1/3 of the original. The performance and temporal consistency can be also significantly stregthened. By using the proposed texture-aware video token merging, the temporal correlation between frames can be further utilized to improve the temporal consistency. We futher propose the referenced self-attention and referenced token merging to support image reference. Experimental results demonstrate the superiority of the proposed method in restoring and enhancing temporally consistent videos.

Modern, neural-based ASR and TTS systems for speech communication are now very sophisticated, but they are frequently designed for average users in more idealized scenarios, which can underperform when interacting with users having specialized listening needs, or who speak in non-standard ways. Here we review recent TTS research, and note that industrial, production-level implementations of adaptive TTS systems are scarce. We describe how we implemented changes to TTS systems based on prior research, and then report qualitative results from two case studies, where adapted TTS techniques are applied to real-world HRI settings: First, in an older adult care home, and second, in an airport, where noise levels create comprehension problems. Results suggest promising directions using relatively simple modifications to pitch, duration, and intensity of the robot's speech, and offer a framework for larger, more formal studies of how TTS adaptations can improve HRI.

Why do some of the world's most powerful Artificial Intelligence (AI) systems still struggle to reason about cause and effect? Causal discovery, the ability to discern the underlying drivers of a system, is fundamental to scientific inquiry and has traditionally relied on structured data and statistical methods. Recently, Large Language Models (LLMs) have been explored as an alternative paradigm for causal inference, leveraging their broad knowledge and reasoning capabilities. Yet, despite their linguistic fluency, LLMs often rely on surface-level pattern recognition rather than genuine causal logic. While it is known that LLMs stumble in this domain, we lack a precise map of where, why, and how significantly these failures occur. In this paper, we propose a systematic stress-test framework across six critical dimensions: model scale, causal graph complexity, prompting strategy, metadata quality, data quality, and sensitivity to uncertainty. Our findings reveal a significant "optimism bias" in LLMs; models consistently overestimate causal links, leading to low precision across the board. Furthermore, we demonstrate that traditional evaluation metrics can be deceptive: while reasoning-heavy models lead in performance, the highest scores often occur when uncertainty in inference is overlooked. Once uncertainty is integrated into the evaluation, performance drops, providing a more honest reflection of causal reliability. We also find that, while scaling model size improves outcomes, these gains are fragile; performance degrades sharply as the quality of metadata decreases. This work serves as a practical guide to the blind spots of LLMs, offering a clear-eyed assessment of when these tools can be integrated into a causal discovery pipeline and when they are likely to lead researchers astray.

Language models often spread a concept across many internal dimensions, making their features difficult to interpret. Sparse autoencoders (SAEs) recover clearer features after training. We ask whether models can instead learn interpretable output coordinates directly. We introduce Mixture of Wide Experts (MoWE), a sparse feed-forward layer in which selected experts read the full representation but write to fixed, disjoint output coordinates. On a controlled task with known hidden causes, this design preserves predictive performance while localizing causes more clearly than dense and standard mixture-of-experts (MoE) models. At approximately matched compute, we train two 19-layer runs of each architecture for 10 billion tokens. MoWE achieves evaluation cross-entropy 2.417 versus 2.452 for dense, while standard MoE reaches 2.22. We measure native-coordinate readability by asking a language-model judge to describe each coordinate from passages where it is strongly active and test that description on new passages. Correcting for descriptions that also fit other coordinates gives a specificity score of 0.523 for MoWE, 0.375 for dense, and 0.268 for standard MoE. TopK SAEs trained on the same FFN updates score 0.489 for dense and 0.556 for MoWE, making native MoWE comparable to the dense SAE. In the first case study, we identify a four-coordinate causal path supporting the model's prediction of degrees after the prompt A right angle is 90. We then copy seven angle-related coordinate values into the model as it processes A rectangle has four, changing its next-token prediction from sides to degrees. In the second, one fixed coordinate set detects hazardous requests and another steers the next token toward refusal. These examples show that MoWE combines sparse conditional computation with coordinates that can be identified and reused directly.

Cardiac auscultation traditionally detects the presence of a heart murmur, but clinicians further characterize it by timing, shape, pitch, quality, and grade. Most automatic phonocardiogram analysis systems stop at binary detection and do not estimate this richer description. In this work, we formulate automatic murmur characterization as five patient-level classification tasks and compare three multi-spot strategies on recordings from the aortic, pulmonary, mitral, and tricuspid auscultation locations. First, classical and deep-learning classifiers are trained independently at each location, and predictions are combined via a maximum-severity late-fusion rule. Second, a synchronous multi-input CNN jointly processes one representative segment per location for a single task. Third, a unified multi-input, multi-output CNN predicts all five attributes from one shared multi-spot representation. Using 735 patients from the CirCor DigiScope dataset with complete four-spot recordings, the best late-fusion configurations reach an unweighted mean sensitivity between 85.56% and 93.81% across the five tasks, with class-specific recalls between 80.00% and 100.00% for every relevant category. The multi-input CNN trails late fusion by 7-16 points depending on the task, while the multi-output model reaches values between 72.59% and 79.40%, estimating all five attributes from one shared representation. We further show that naively balancing training data by adding all minority-class segments degrades patient-level performance under the maximum-severity rule, since it introduces non-representative, noisy segments not filtered by the clustering-based segment selection used for evaluation. These results indicate multi-spot murmur characterization is feasible with both fusion-based and jointly learned architectures, outlining a path toward automatic, structured murmur reports complementing binary screening.

Bubeck and Sellke's universal law of robustness says that fitting $n$ noisy samples below the noise level with a class of $p$ real parameters forces a Lipschitz constant of order $\sqrt{nd/p}$, when the covariate distribution on $\mathbb{R}^d$ is isoperimetric: smooth interpolation costs a factor $d$ of overparametrization. Its hypotheses are Gaussian-type concentration, a homogeneous covariate law, scalar outputs, and independent samples; we relax each one. Exponential concentration suffices, at the unchanged rate and absolute constant and at the price of a capacity burn-in, some form of which we show is necessary; this gives the law for normalized log-concave covariates, with a $\log d$ loss from the current KLS bound. For heterogeneous mixtures the worst component constant is replaced by the weighted average of the component constants, under a typicality condition on the mixture composition. For vector-valued predictors the bound at the Frobenius Lipschitz scale is free of the output dimension $K$, and the Euclidean bound loses only $K^{-1/2}$, a loss that covariate-independent noise removes. Stationary Markov covariates retain the i.i.d. rate, the pseudo-spectral gap entering only the sample-size conditions. One master theorem carries the shared argument, and constants are explicit throughout.

We study short-horizon neonatal seizure prediction on Helsinki-style (HUH) preprocessed EEG feature tables using quantum-inspired reservoir computing. Under strict subject-wise cross-validation on imbalanced preictal–interictal windows, we test whether reservoir constructions shaped after (i) gate-based qubit dynamics and (ii) continuous-variable (bosonic/photonic) dynamics provide a practical sequence encoder relative to strong classical baselines. Both reservoirs are executed here purely in classical simulation on a single GPU; accordingly, we do not claim quantum advantage, but rather evaluate whether quantum-shaped dynamics act as useful, low-footprint random feature maps with a trained linear readout. We compare against Random Forests and RBF–SVMs, and report AUROC, AUPRC, per-subject AUPRC, expected calibration error, and stratified bootstrap confidence intervals. Across our experiments, the CV/bosonic reservoir is consistently competitive and can improve AUPRC over the baselines, while the gate-style reservoir is more sensitive to hyperparameters. These results suggest that quantum-inspired, photonics-motivated reservoirs can be practical sequence encoders for neonatal EEG feature tables under conservative subject-wise evaluation.

In e-commerce advertising, ad platforms typically fail to model the externalities among advertisers when making optimization decisions: an advertiser's actual performance is largely shaped by rivals competing for the same users; these rivals typically operate in overlapping product categories and at comparable budget scales. However, capturing this rivalry as an explicit, reusable signal is challenging because it is directed, multi-faceted, and partially observable. Existing paradigms such as interaction graphs, taxonomy rules, and single-embedding models do not fully capture these properties. In this paper, we propose LACRM (LLM-driven Advertiser Competition Relationship Mining), a novel framework that constructs directed competition graphs with continuous edge intensities via a three-stage pipeline. LACRM moves beyond symmetric similarity through LLM-driven role-separated profiling, which enables asymmetric multi-route retrieval to uncover both realized and latent rivals. The framework further incorporates listwise learning-to-rank aligned with temporal auction behavior to accurately model directed competition intensity. In offline retrieval experiments on an industrial benchmark, LACRM outperforms strong directed GNN baselines and ablation variants. Multi-LLM and human studies provide complementary evidence that the recovered latent rivals reflect meaningful market competition rather than merely model-generated artifacts. In production A/B tests, LACRM-guided bidding and incentive allocation increase advertising revenue by 1.73% and 1.14%, respectively. The results demonstrate the practical value of explicit competition modeling in large-scale advertising systems.

We present a focused and reproducible study of multi-horizon forecasting on the Lomnicky Stit neutron monitor (LMKS) time series. Our evaluation suite covers simple seasonal baselines, modern deep sequence models, and functional and quantum-inspired architectures, including Seasonal Naive, Long Short-Term Memory (LSTM), Temporal Convolutional Network (TCN), N-BEATS, Kolmogorov-Arnold Networks (KAN), and two quantum-inspired variants, QiLSTM and QiKAN. We describe the dataset characteristics, diagnostic analysis, preprocessing pipeline, and training procedures, and report aggregate point-forecast performance using mean absolute error (MAE) and root mean squared error (RMSE) for all evaluated models. Our quick-run results indicate that the quantum-inspired KAN variant, QiKAN, achieves the lowest aggregate forecasting error among the evaluated configurations, while the simple Seasonal Naive baseline remains remarkably competitive. These results suggest that, for highly periodic scientific monitoring time series, models incorporating strong seasonal or low-dimensional functional priors can match or outperform substantially more complex sequence architectures. The findings motivate further investigation of parsimonious and decomposable function approximators for forecasting periodic scientific signals.

Text-to-SQL systems are increasingly deployed on production databases, where queries that pass benchmark evaluation can still produce results that distort downstream workflows. Standard set-based execution accuracy (Set-EX) collapses duplicate rows and can therefore miss multiplicity errors, including missing DISTINCT, inflated aggregates, and Cartesian-style join explosions. We call this the Multiplicity Blind Spot (MBS) and introduce Multiset-EX, a multiplicity-preserving evaluation criterion that exposes such failures. Across released DeepEye-SQL artifacts from three backbones (Qwen2.5-Coder-32B, Qwen3-Coder-30B-A3B, and Gemma-3-27B) on executable BIRD-Dev N=1532, we find a consistent 5.81–6.79 pp gap between Set-EX and Multiset-EX. The gap is not specific to DeepEye-SQL: it persists on released DAIL-SQL+GPT-4 (5.22 pp) and BIRD GPT-3.5-turbo (3.39 pp) predictions. We further introduce ModularSQL, a lightweight post-selection runtime guardrail that probes executed results for multiplicity anomalies and applies deterministic patches or low-cost LLM rescue only to flagged queries. Integrated with DeepEye-SQL using Qwen3-Coder, ModularSQL preserves Set-EX at 72.06% while improving Multiset-EX from 65.86% to 67.75% (+1.89 pp). It flags 77 high-risk anomalies, while adding only $0.0076 in total LLM cost and 120 ms amortized latency per query. Cross-pipeline evaluation shows that the candidate-free detector and deterministic patches also transfer to independently released prediction sets. Overall, these results show that benchmark accuracy does not necessarily imply execution-safe SQL, and that lightweight, multiplicity-aware runtime guardrails can narrow this gap with modest computational overhead.

Quantum reservoir computing (QRC) uses fixed quantum dynamics to encode a time series and trains only a classical readout, but a benchmark capacity alone does not reveal whether task information is lost in the reservoir, the measurement, feature compression, or finite sampling. We introduce task-resolved Fisher spectroscopy, in which prediction targets define orthonormal score coordinates on the stationary distribution of input histories. Reweighting labeled histories along these scores generates an exactly affine family of reservoir states and measurement outcomes. In the same coordinates, we obtain a many-body Fisher-information hierarchy relating the state quantum Fisher information, the Fisher information of the complete measurement record, and moment matrices retained through many-body order $r$. The quadratic form of each moment matrix is exactly the stationary capacity of the optimal linear readout, while a finite-measurement extension predicts the approach from one-shot records to ideal expectation-value features. The measurement-level quantities require only stationary labeled records and measured outcome strings, not an input model or quantum-state tomography. In a five-spin open reservoir, interactions route fourth-order temporal information into higher-body correlations, so low-order compression can incur orders-of-magnitude sampling overhead even when a conventional parity target remains accessible. For correlated inputs and outputs, record-defined task scores predict held-out capacities, the required feature order, and measurement-budget dependence; optimizing the local measurement axis recovers otherwise hidden task information. The framework connects benchmark performance to experimentally actionable diagnoses of reservoir encoding, measurement choice, classical representation, and shot allocation.

We record lower bounds for permanent circuits with valid division over characteristic zero, counting nonscalar multiplications and divisions while additions and scalar operations are free. Chapter 5 of OpenAI's Ten Advances supplies the block construction and critical-locus estimate; these, with the parameter choice made here and the classical bounds of Strassen and Baur-Strassen, give liminf as n tends to infinity of L_div(per_n)/(n^2 log_2 log_2 n) >= 1/12. Our finite-parameter refinement for matching-minor polynomials observes that the coefficient vector at deletion size h lies in a space of dimension at most min{binom(t,h), binom(t,d-h)}. Twelve machine-checked declarations of the Lean development accompanying OpenAI's manuscript on border determinantal complexity of the permanent (September 24, 2026) imply, by a short written argument, the existence of a geometric slice; a further written, unformalized application of the same criterion gives L_div(per_n) >= (n^2/119790) log_2(n/44) for n >= 1936. Assuming Proposition 7.3 of that manuscript gives the bound (n^2/86400) log_2(n/48) for n >= 1408. These are order-of-growth results; the elementary Hessian bound n^2/2 is larger at practical orders.

The aim of this work is to provide a rigorous mathematical analysis of a well-known SIS epidemic model with a saturating contact rate. First, we establish the global asymptotic stability (GAS) of the model's equilibria by employing a suitable Lyapunov function in combination with the Poincaré–Bendixson theorem and the Bendixson–Dulac criterion. The resulting GAS results improve upon previous findings for this SIS model and may also be applicable to its extensions with more general saturating contact rates. Second, we construct families of first- and second-order nonstandard finite difference (NSFD) schemes that preserve the positivity and asymptotic stability properties of the SIS model for arbitrary step sizes. All these schemes are formulated within Mickens' framework; however, compared with their first-order counterparts, the second-order schemes employ a more elaborate construction that combines a weighted nonlocal approximation of the right-hand side with suitably renormalized denominator functions. The weighted nonlocal approximation guarantees dynamic consistency, whereas the denominator functions ensure second-order convergence. Finally, we conduct numerical experiments to validate the theoretical results and demonstrate the advantages of the proposed NSFD schemes. The numerical results are in good agreement with the theoretical results. The approach developed in this work is applicable not only to other epidemiological systems but also, more generally, to mathematical models arising in a wide range of real-world applications.

Detecting and classifying Android malware families remains challenging due to high feature dimensionality, class imbalance, and the high cost of expert-labeled data. Semi-supervised learning (SSL) offers a way to leverage unlabeled samples, but prior works rarely test whether SSL benefits generalize across classifier types or report statistical significance. We present a systematic evaluation of pseudo-labeling across six classifiers (LightGBM, XGBoost, Random Forest, Logistic Regression, MLP, and SVM) on the CICMalDroid 2020 dataset, using five-fold stratified cross-validation and paired t-tests across five labeled ratios (1-20%). We find that SSL benefit is strongly classifier-dependent: SVM shows the largest significant gain (+4.4% accuracy at 5% labels, p = 0.0028), LightGBM improves modestly (+0.8 to +1.3% at 2-5% labels), while Random Forest is significantly harmed at low label ratios (-3.1% at 1% labels). Per-class analysis reveals SSL disproportionately benefits the hardest-to-classify families, with Adware F1 improving by +13.8 percentage points versus only +0.8 for the already well-classified Benign class. We further show that approximately 800 labeled samples (10% of the dataset) yield near-optimal performance across all classifiers. These findings offer practical guidance on when and with which classifier pseudo-labeling is worthwhile for Android malware classification.

Reasoning-enhanced universal multimodal embeddings (UME) improve heterogeneous retrieval, but plausible rationales do not necessarily produce discriminative rankings. We study this gap by comparing the discriminative (DISC) and reasoning-driven generative (GEN) branches of UME-R1, a state-of-the-art reasoning UME method. We decompose reasoning utility into positive-target gain, hard-negative gain, and their margin difference. Positive similarity increases for 56.6%, but 15.7% are false-helpful cases where reasoning moves hard negatives closer even more. Local-neighborhood and token-attribution diagnostics suggest why: reasoning often de-condenses retrieved neighborhoods, but utility requires separator-aligned movement, while influential CoT tokens frequently encode evidence shared by positives and hard negatives. Motivated by these diagnostics, we propose SURE (Score-structure Utility Router for Embeddings), which improves UME-R1-7B by 1.5 points and yields consistent gains on two additional embedding models on MMEB-V2, without retraining, label-based policy selection, or extra VLM forward passes.

Arabic-Russian machine translation (MT) remains under-explored due to the rich morphology of Arabic and low lexical overlap between the two languages. We benchmark seven fine-tuned neural machine translation (NMT) models against four few-shot large language models (LLMs) on a 20k/5k/5k split of a new 15.47M-pair corpus. Fine-tuned NLLB-1.3B achieves the highest BLEU (16.3) and COMET (0.738). Aya-Expanse 8B leads the few-shot LLMs (BLEU 1.7 on 500 sentences, chrF 25.7), but all LLM scores remain far below the fine-tuned NMT baselines. Error analysis identifies low lexical overlap as the dominant failure mode; among the worst translations, mT5-small produces 32% too-short outputs. Bootstrap tests confirm significant differences among most models. Our results demonstrate that fine-tuned NMT significantly outperforms few-shot LLMs for Arabic-Russian translation under low-resource conditions.

Navigation World Models (NWMs) predict action-conditioned visual futures for planning. Two practical challenges are central to their design: selecting a suitable visual representation and efficiently modeling observation history for repeated candidate queries. Standard Global-Softmax attention provides flexible interactions but repeatedly processes the same history, leading to increasing computation and memory costs for long contexts and multi-query planning. We study both problems within a unified conditional flow-transformer framework. We first compare five frozen visual representations under the same dynamics model and evaluation. To reduce redundant history computation, we design Cached-Linear, a hybrid architecture that combines local and shifted-window attention for target mixing with linear attention for reusable history access. We further develop Balanced Gated Delta Network (GDN), which augments this design with frame-wise recurrent memory for temporal history modeling. Experiments on RECON, SACSoN, and SCAND show that representation choice depends on the prediction objective: PAE-L performs best for reconstruction, RAE-B for direct prediction, and V-JEPA for long-horizon rollout. Under shared-history workloads, Cached-Linear substantially reduces computation and memory compared with Global-Softmax, while Balanced GDN improves selected direct-prediction endpoints with efficient context reuse. Overall, we systematically study visual representation and history modeling for NWMs and develop hybrid reusable-history architectures for efficient long-context and multi-query prediction.

Automatic stress recognition from facial video provides a non-contact approach for affective monitoring. However, most existing video-based methods divide complete recordings into shorter temporal segments before performing classification. Such segmentation requires additional decisions concerning segment duration, overlap, and prediction aggregation, and may restrict the model from exploiting information distributed across the entire recording. We introduce UNWIND, a facial-video framework for stress detection that analyzes a complete recording as a single model input, eliminating the need for temporal windowing or external segmentation. UNWIND reorganizes the video by folding its temporal dimension into the channel dimension of a two-dimensional spatial representation, which is subsequently processed through a unified asymmetric-attention architecture. With a temporal stride of $\tau=1$, the framework processes the entire $120$-second sequence, corresponding to $3{,}600$ frames sampled at $30$~fps, in a single input. We evaluate seven temporal-stride settings on a stress dataset comprising $58$ subjects, using a stratified subject-level protocol that covers configurations from dense frame retention to sparse temporal sampling. The highest test accuracy, $70.02\%$, is obtained at $\tau=15$, while processing all frames at $\tau=1$ achieves a comparable accuracy of $69.73\%$. Computational requirements range from $12.48$ to $348.78$ GFLOPs across the evaluated stride settings, illustrating the balance between temporal sampling density and computational efficiency. The findings show that effective facial-video stress recognition can be achieved without dividing recordings into temporal windows and that complete-recording inference can be performed within a single unified model.

Sparse autoencoders (SAEs) decompose LLM activations into sparse dictionary atoms, so that each distinct concept gets its own feature. One recurring behavior complicates this premise: feature absorption, in which a parent concept and its children–fruit and {apple, banana, pear}, say–collapse into a shared family direction. Prior work documents absorption empirically; missing is a closed-form prediction of when the shared direction is the cost-optimal representation of an active semantic family. This paper closes that gap. For a hierarchical Bernoulli generator with $k$ active children and residual scale $\alpha$, the $L_0$-penalized reconstruction objective admits a closed-form phase boundary $\lambda_c(k,\alpha)=\alpha^2 k/(k-1)$: above it, pure parent absorption is strictly cheaper than pure child coding. Building on this boundary, we introduce HiPACE, an evaluation protocol that tests the boundary's structural consequence in real SAE dictionaries–measuring parent–child decoder structure over WordNet families, freezing the discovery-selected statistic before testing on unseen families, and contrasting genuine families against randomized sibling nulls. The boundary proves sharp in its native regime, predicting the synthetic transition within $\pm15%$ on all 30 tested cells. In Pythia-160m SAEs, the parent–child decoder gap recovers the predicted ordering with partial correlations up to $-0.93$ that sustain on the locked holdout and exclude sibling nulls ($p=0.002$). Controlled activation composition connects the theory's active-child count to the recovered family directions, and residual-stream interventions show that signed family directions increase parent-category logits, reversing under sign flip and vanishing under random controls–establishing causal sufficiency at the family-subspace level.