Search Results (763)

Background: Cryptocurrency allocation tools that use graph learning are often judged on gross performance and rarely stress-tested under realistic trading frictions. Methods: We evaluate a regime-supervised graph-neural tangency pipeline on 33 large-cap Binance USDT pairs (April 2021–June 2026), combining rolling Pearson correlation graphs, GCN/GraphSAGE/GAT inference under a composite Sharpe–modularity–ranking loss, and a hidden Markov layer that gates graph sparsity and long-only constraints. Six classical benchmarks share the same five-day calendar and ten-basis-point costs. An in-sample Kupiec coverage check is reported only as a diagnostic identity and is not treated as out-of-sample VaR validation. The contribution is an empirical stress test under trading frictions, not a new learning architecture: unconstrained dynamic graphs destroy net performance after costs; on this sample, a matched hidden-Markov supervisor records a higher net Sharpe ratio than a volatility-threshold rule, a descriptive difference that is not tested for significance; and no graph specification ranks above passive benchmarks on full-sample headline metrics. Results: Unconstrained dynamic graphs raise turnover by about 67% and produce negative net Sharpe; static GCN records the best GNN Sharpe (0.31), while HMM-gated GraphSAGE recovers a positive dynamic Sharpe (0.18). On this sample, cap-weight and minimum-variance rules rank above every GNN specification on headline Sharpe and drawdown; in the ablation, the HMM supervisor records a net Sharpe of 0.18 against $-0.09$ for a matched volatility-threshold rule on this sample, with no formal significance test. Conclusions: Graph adaptivity is not free. Regime supervision is associated with stronger dynamic performance in the 2023 recovery and 2024 bull windows and with weaker performance in 2022 and 2025–26. The evidence supports a sample-specific risk-policy evaluation, not an unconditional alpha claim or a practically viable allocator: positive Sharpe ratios coexist with maximum drawdowns of $-82\%$ to $-99\%$. Full article

Threshold-based flight data monitoring (FDM) identifies parameter exceedances but provides limited visibility into gradual operational deviation within nominal operating bounds. This study presents a three-layer framework for continuous operational deviation quantification from quick access recorder (QAR) data. A standard operating procedure (SOP)-constrained Gaussian mixture model–hidden Markov model (GMM-HMM) infers 27 procedural states from flight parameters, aligning latent state representation with operational procedure. From these state sequences, maximum entropy inverse reinforcement learning (IRL) recovers an expert value function that estimates execution quality at each time step. Split conformal prediction calibrates detection thresholds with distribution-free coverage guarantees, while counterfactual analysis traces identified deviations to their procedural origin. Learned from real flight data, the recovered reward function assigns 1.6 times greater weight to lateral and speed stability than to vertical stability. Split conformal calibration yields an empirical false alarm rate of 4.2% under a nominal 5% significance level. Counterfactual analysis identifies vertical speed management as the dominant source of procedural deviation. These results support the proposed framework as a candidate continuous safety performance indicator that surfaces gradual operational deviation that threshold-based exceedance monitoring would not flag. Full article

Understanding how visual exploration evolves over time during complex decision making remains challenging, as the dynamics of underlying cognitive processes are difficult to observe. Hidden Markov models are widely used to model eye movement sequences, but their assumption of stationary transition probabilities limits their ability to identify successive phases. This study investigated whether Switching Hidden Markov Models (SHMMs), trained only on fixation positions, could segment visual exploration during an obstacle-avoidance task requiring a left/right decision. Individual SHMMs with two or three high-level (HI) states were trained without using reaction times. HI states were characterized through transition-matrix variability, saliency maps, cumulative dwell times in regions of interest, and their association with reaction times. Both models produced temporal segmentations consistent with decision making. Early HI states showed greater interindividual variability and exploration focused on obstacles, whereas later states were more homogeneous and reflected decision confirmation followed by reorientation toward the center of the screen. The transition to the final HI state of the three-state model best explained reaction time. Furthermore, the average log-likelihood of the Viterbi state restoration improved the fit quality of this explanatory model, indicating that SHMM confidence provided complementary behavioral information. These findings suggest that SHMMs trained only on fixation positions can identify phases broadly consistent with those traditionally observed in the decision-making process, such as early evidence accumulation and late decision confirmation. Full article

Cost settlement for temporary repair of ship equipment is characterized by lengthy ex post audits and the lack of a directly quotable pricing benchmark. Given the intertwined effects of tight schedules, holiday wage premiums, and fluctuating resource availability—and the consequent need for historical memory and foresight in the model—an LSTM–PPO comprehensive unit price estimation model is developed. Because a standard Markov decision process cannot distinguish different historical paths or exploit forward-looking information, the problem is formulated as a finite-horizon partially observable Markov decision process (FH-POMDP), with a corresponding observation vector and a composite reward function. To capture time-varying holiday rates and the path dependence of historical trajectories, an LSTM encodes the full observation sequence and, through its gating mechanism, fuses historical trajectories with temporal changes in holiday windows in the hidden state, enabling the policy to anticipate rate shifts and allocate labor input in advance. For the hybrid action space of daily mode selection and intensity adjustment, hybrid entropy regularization is introduced to discourage premature collapse onto a single mode preference early in training and to improve robustness across diverse scenarios. Experiments show that the proposed method produces benchmark unit prices with smaller deviations from actual settlement prices than the baselines on the test set, and that it can discriminate holiday windows with different rate multipliers and resource conditions, thereby providing technical support for the ex post settlement of emergency support funds. Full article

Static e-commerce segmentation leaves category change in payment data unmeasured. We analyze 96.19 million credit-card transactions, including online-shopping categories but no channel flags, across 13 quarters for 229,586 customers, measuring transition exposure, set turnover, retention, and novelty. Temporal homogeneity was rejected (${\chi }^2\left(132\right)=\mathrm{57,222.2}$), and a second-order Markov model outperformed a first-order model in held-out log loss (0.886 vs. 0.913). The hidden Markov benchmark matched within ±0.005 log loss, but only two of five fold intervals excluded zero; the nonhomogeneous specification did not improve. A pre-registered audit found a mean-spending ΔR² of 0.0009 over the original baseline after recency correction; neither three primary decision targets nor the auxiliary adoption target reached the area under the receiver operating characteristic curve (AUC) threshold of ΔAUC 0.01 in any fold. One of three gates passed; its increment fell below the residual criterion with lagged category-change information. The re-analysis added nonlinear learners, recent-window and monthly resolutions, and spending-change and category-entropy targets; all increments stayed below 0.01. Equivalence tests placed 95% upper bounds at 0.0038, 0.0017, and 0.0032, bounding increments below materiality. We contribute an exposure-aware audit reporting this bound on payment-category data for e-commerce analytics. Full article

Supply chain plasticity (SCP) is a critical dynamic capacity through which firms respond to disruptions by reconfiguring their supply chains. Its latent, regime-dependent nature makes SCP difficult to identify. Grounded in a social–ecological systems view, this study aims to infer SCP endogenously from longitudinal supply chain networks by developing a regime-conditioned probabilistic framework (RCPF) that integrates graph-theoretic measures with latent-variable models. The framework enables the endogenous inference of regime-transition states, the tracing of firm-level SCP dynamics, and the identification of cross-firm SCP archetypes through three sequential probabilistic modules. Specifically, a hidden Markov model is specified to decode the latent regime-transition-state trajectory from the graph edit distance and spectral distance between consecutive network snapshots. Next, a regime-conditioned hidden Markov model is formulated to trace firms’ SCP dynamics from local relational adjustments and network positional variations, with latent-state transitions conditioned on the posterior distribution over the inferred regime-transition states. Finally, a finite mixture model is applied to identify interpretable SCP archetypes from phase-specific profiles constructed from firms’ posterior plasticity-state probabilities across regime-shift phases. Applied to China’s electric vehicle supply chain network, the framework identifies regime shifts aligned with disruptive developments and shows that regime-shift conditions increase the likelihood and persistence of firms’ structural reconfiguration. The inferred SCP archetypes reveal role-specific adaptation pathways and resilience outcomes, informing firms’ differentiated strategic responses to disruptions. Full article

Background/Objectives: Antimicrobial peptides (AMPs) contribute to plant innate immunity and abiotic-stress adaptation, but their repertoires remain poorly characterized in extremophytic plants. We aimed to define AMP gene families and their expression under Na₂SO₄ treatment in the xerophytic recretohalophyte Reaumuria soongarica. Methods: Six canonical AMP families were screened using hidden Markov model and homology searches, domain verification, gene-model and transcript-support assessment, nsLTP boundary curation, phylogenetic and synteny analyses, and reanalysis of public RNA-sequencing data from 0, 200, and 400 mM Na₂SO₄ treatments. Results: We identified 73 candidates: 15 defensins, 18 Snakin/GASA genes and 40 nsLTPs, including seven previously unannotated, transcript-supported models. RsLTP20 was reported separately as a reference-dependent C4-like candidate. Four sequence and alignment strategies supported two defensin, one Snakin/GASA and three nsLTP local groups. No tested promoter pattern was significantly enriched after multiple-testing correction. At 400 mM Na₂SO₄ relative to the control, 18 of the 66 catalogue members represented in the quantification reference were differentially expressed (12 upregulated and six downregulated; BH-adjusted p < 0.05 and |log₂ fold change| > 1). Comparison with 38 Tamarix austromongolica candidates identified 27 reciprocal-best-hit pairs; genomic collinearity supported 25 direct candidate pairs. A separate Arabidopsis comparison recovered 26 candidate-anchored pairs. Conclusions: This curated catalogue defines conservative phylogenetic groups and provides candidates for experimental analysis of stress adaptation and peptide function. Full article

Background/Objectives: Vesicular trafficking mediates the transport of proteins and other cellular components between intracellular organelles. The MAG2 complex serves as a key tethering factor mediating ER–Golgi retrograde vesicle transport, and has been implicated in plant growth, development, and stress responses. Methods: In this study, seven MAG2 family genes were identified in Medicago sativa from the Zhongmu No. 4 reference genome using hidden Markov model searches (HMM) and BLASTP analysis. Results: Phylogenetic analysis assigned four genes to the MAG2 subfamily and three genes to the MAG2L subfamily. Chromosomal distribution and collinearity analyses indicated that segmental duplication potentially contributed to the expansion of the MAG2 family in alfalfa. Gene structure and conserved motif analyses revealed a high degree of conservation among the identified members, with Motif 10 occurring specifically in the MAG2 subfamily. Subcellular localization prediction analysis predicted that four MsMAG2 proteins were localized in chloroplasts, two in the nucleus, and one in the peroxisome. Analysis of promoter sequences identified numerous cis-regulatory elements associated with responses to hormones, light, and environmental stress, suggesting that MsMAG2 genes may participate in plant development and environmental adaptation. Furthermore, genome-wide association analysis identified MsMAG2L2 that can be associated with tillering. Conclusions: These results provide a comprehensive genome-wide characterization of the MAG2 gene family in alfalfa and establish a basis for further investigating the functional associations of MsMAG2L2 in branch development. Full article

Sequential object-oriented software remains difficult to parallelize because several important decisions are not entirely local. A useful decomposition should consider architectural structure, execution frequency, communication latency, and how much communication can overlap with local computation. In contrast, many traditional modularization methods focus on cohesion, coupling, or communication cost without explicitly considering whether an unavoidable remote call can be issued early enough for part of its latency to be hidden by independent work. DTMC-GA addresses this problem by combining a Discrete-Time Markov Chain (DTMC) execution model, a Distributive Execution Time (DET) estimator, and a Genetic Algorithm (GA) search procedure. Guarded call-use displacement enlarges modeled overlap windows only when the relevant dependence, ordering, and safety conditions are satisfied. The public artifact contains five local benchmark fixtures, five comparison methods, 30 retained runs per algorithm–fixture pair, raw outputs, statistical scripts, regenerated figures, Docker support, continuous-integration verification, and a Zenodo DOI. Within this released five-method panel, DTMC-GA achieved the highest aggregate mean modeled speedup of 2.708 ± 0.329. A more recent NSGA-III comparator was also evaluated using matched fixtures, DTMC–DET inputs, constraints, random seeds, population size, and generation budget. NSGA-III achieved an aggregate mean modeled speedup of 2.737 ± 0.323, 1.07% above DTMC-GA, although none of the five fixture-level paired differences remained significant after Bonferroni correction. A separate sparse-graph scalability study covered 50, 100, 250, 500, 750, and 1000 classes; mean analysis time increased from about 0.024 to 0.286 s, while peak memory rose from about 119.1 to 120.9 MiB. The study also includes an executed OpenCV 4.7.0 calibration using a fixed five-convolution DNN workload. Across 30 matched pairs, native-parallel OpenCV achieved a measured ratio-of-means speedup of 1.8691×. The bounded empirical DTMC-GA/DET rerun over 30 seeds produced a mean modeled speedup of 1.0000× with an SD of 0.0000, giving an absolute modeled-versus-measured difference of 46.50%. Because the measured parallel comparator is OpenCV’s native threaded implementation rather than a DTMC-GA-generated source rewrite, this experiment is interpreted as real-runtime model calibration rather than direct measured DTMC-GA optimized speedup. Full article

Fatty acid desaturase (FAD) is a core catalytic enzyme in plants for the synthesis of unsaturated fatty acids, profoundly affecting plant growth, development, and adaptability to various environmental stresses. The walnut (Juglans regia L.) is an important woody oil tree species, and its kernel is rich in unsaturated fatty acids. Systematic identification of the walnut FAD gene family and analysis of its function are of great significance for revealing the molecular mechanisms underlying unsaturated fatty acid metabolism in the walnut. Based on walnut whole-genome data, this study used homology alignment and hidden Markov model search methods to identify the JrFAD gene family members. Subsequently, a variety of bioinformatics tools were used to systematically analyze their structural characteristics, evolutionary expansion mechanism, expression regulation, and function. A total of 21 JrFAD gene family members were identified and classified into five subfamilies. The family genes were unevenly distributed on nine chromosomes. WGD/segmental duplication was the main expansion method, and the duplicated gene pairs experienced strong purification selection. The family gene promoter sequence is rich in regulatory elements that respond to light, plant hormones, and various stresses. The expression pattern analysis showed that JrFAD3.1 and JrFAD2.3 showed high expression specifically during the rapid accumulation of walnut kernel oil. This study clarified the composition and evolutionary characteristics of the FAD gene family in the walnut, which provides useful information for in-depth analyses of its functional mechanism in the regulation of lipid metabolism, and also identified potential candidate gene resources for the genetic improvement of walnut varieties with high amounts of unsaturated fatty acids. Full article

Bitcoin option prices reflect terminal variance and the cost of managing convex exposure in a market with changing depth and execution quality. This paper asks whether a liquidity state can be separated from fractional rough volatility in Bitcoin option valuation. The contribution is a modelling combination: standard stochastic-calculus and rough-volatility tools are joined to a regime-switching hedging-cost reserve, producing a leading-order at-the-money implied-volatility lift. The empirical design tests a liquidity–IV association and its scale using a 700-contract Deribit snapshot, a 4513-trade 24-h window spanning two UTC dates, a 775,315-trade panel over 92 dates, Ether replication, and placebos. The association is strong in open-interest-weighted specifications and for puts, but is absent for calls; it remains after controlling for option premium. Leave-one-expiry-out validation improves open-interest-weighted RMSE but not unweighted RMSE. A realised-volatility HMM is only a market-stress diagnostic, not an estimated liquidity regime. An empirical one-step hedging exercise does not validate the model’s simulated hedging comparative static. Accordingly, the evidence is associational, put-side, and narrower than a causal or fully structural validation. Full article

Towed hydrophone arrays are widely employed for underwater target detection and direction-of-arrival (DOA) estimation. However, array shape distortion induced by ocean currents, internal waves, and platform maneuvers severely degrades beamforming performance and DOA estimation accuracy. In this paper, a novel Bayesian fusion framework is proposed to achieve robust array shape estimation. Specifically, based on the time-delay estimates derived from the phase differences of line-spectrum components in a pre-processing step, the array geometry is first reconstructed via a piecewise straight-line fitting method. Concurrently, an existing hidden Markov model (HMM)-based method is adopted to estimate the inter-segment deviation angles, in which the smoothness of the array shape is enforced through the state-transition probabilities. The proposed framework then treats these two preliminary estimates as observations from distinct sources and incorporates a smoothness prior within a maximum a posteriori (MAP) formulation that admits a non-iterative closed-form solution to enforce physical continuity constraints on the array geometry. By fusing these complementary estimates, the proposed method simultaneously preserves local sensitivity to fine-scale bends and maintains global consistency of the array shape. Both simulation and lake-trial experiments validate the effectiveness of the proposed method, reducing the array shape estimation error by more than 30% relative to representative existing methods. Moreover, by relying solely on the received acoustic data, the method lowers the dependence on auxiliary sensors and the associated system cost. Full article

With the large-scale deployment of smart electricity meters in China, power companies need online methods that can evaluate meter operating status and locate faulty units without field inspection. Meter data are inherently multimodal. They combine time-series measurements with textual event logs. The semantic gap between these modalities limits the accuracy of existing approaches. This paper proposes an online operational status evaluation method for smart meters based on cross-modal perception with large language models. A Bi-LSTM and TCN-Attention network with quantile regression first builds a robust district line-loss baseline that captures seasonal fluctuations and operational uncertainty. A cross-modal alignment module then fuses the two modalities. The measurement sequences are encoded by PatchTST, and the event logs are encoded by a LoRA-fine-tuned LLM. The fusion is performed through contrastive learning and gated fusion. A retrieval-augmented knowledge graph provides additional support. Finally, a multi-indicator health index grades meters into five condition levels, and a hidden Markov model estimates the remaining useful life. Case study results demonstrate that the proposed method achieves 85.7% accuracy, outperforming the strongest baseline. Full article

Unprotected left turns at urban intersections require drivers to continually regulate their driving behavior while negotiating conflicts with opposing through traffic. Existing studies have mostly examined such maneuvers through gap-acceptance decisions and surrogate conflict indicators, which cannot reflect how driving behavior changes for unprotected left-turn interaction events. To explore the evolution of unprotected left-turn interactions, this study develops a data-driven framework to decompose continuous left-turn trajectories into interpretable, variable-length driving primitives. Using high-resolution unmanned aerial vehicle trajectory data, 2490 valid left-turning and opposing-through vehicle interaction events were extracted. A Non-Homogeneous Hidden Markov Model was adopted to segment each interaction event into driving primitives. These primitives are then clustered using Time-Series K-Means with Dynamic Time Warping. The clustering results yielded six behavior patterns: cautious turning, low-speed waiting, accelerating departure, deceleration observation, intensive turning, and steady driving. These patterns were then mapped back to the temporal sequence of each interaction event to construct behavioral transition chains. The results demonstrate that an unprotected left turn typically evolves as an ordered combination of the identified behavior patterns, from waiting and observation to turning and departure, and that transitions among these patterns are associated with changing opposing-traffic conditions. Full article

Addressing multi-factor coupling, progressive degradation, and time-varying operating conditions in substation equipment under high-penetration renewable energy integration, this paper proposes a Time-varying Hidden Markov Model (T-HMM) for transformer fault diagnosis using multi-source heterogeneous data. Unlike conventional HMM with fixed transition matrices and initial parameter sensitivity, the proposed framework introduces a forgetting-factor-driven online transition matrix updating mechanism, enabling adaptive state tracking under varying operating conditions. A multi-dimensional feature system is constructed incorporating fundamental state, coupling correlation, and temporal evolution characteristics, with state-dependent sliding window standardization and adaptive wavelet denoising to enhance early-stage fault discriminability. Parameter optimization employs pre-clustering and multi-start strategies to circumvent local optima in the Baum–Welch algorithm, while an improved decoding strategy achieves real-time health state identification and multi-step probability prediction. Experimental results demonstrate that the proposed method accurately tracks state evolution trends and effectively identifies fault categories, achieving significantly superior state recognition accuracy and multi-step prediction hit rates compared to conventional HMM, with substantially reduced mean absolute error. The method effectively captures subtle precursors during normal-to-fault transitions, providing reliable theoretical foundations for early fault warning and condition-based maintenance. Full article