Search Results (201)

The utilization of photovoltaic (PV) energy has witnessed significant recent growth, positioning it as the fastest-growing renewable energy technology. This rise in deployment underscores the importance of monitoring PV systems, which has become an increasingly critical area of research. However, the heightened focus on monitoring has also revealed a key challenge: various types of faults often remain undetected, potentially causing severe performance degradation. This study proposes a machine learning-based fault classification method that integrates Principal Component Analysis (PCA) for feature extraction with Kullback–Leibler (KL) divergence for distribution-based classification. The approach is applied to a publicly available dataset on a single benchmark of a 5 kW PV plant containing voltage, current, temperature, and irradiance measurements for five operating conditions: normal operation, short circuit, open circuit, partial shading, and degradation. A dedicated normalization strategy and optimal bin-width selection are employed to enhance the stability and accuracy of probability density function estimation. Experimental results demonstrate that the proposed PCA–KLD framework achieves superior accuracy compared to existing methods applied to the same dataset. These results confirm the effectiveness and robustness of the approach for fault classification in PV systems and highlight its potential for handling challenging fault scenarios. Full article

As social infrastructure intensively developed during the high economic growth period of the 1970s faces simultaneous aging, there is an urgent need to transition from conventional reactive maintenance to preventive maintenance utilizing various data (data-driven asset management. However, the greatest barrier in practice is that inspection data is unevenly distributed in analog formats such as paper and unstructured files, and heavily relies on the subjective visual evaluation of expert engineers (e.g., discrete graded evaluations from A to D). The intervention of this “Assessor Bias” makes it difficult to ensure the robustness required for direct statistical analysis. This paper serves as a bridge between this analog expert knowledge and quantitative data science. It formulates human cognitive conflicts (true state, peer pressure, avoidance of cognitive load) using the distance-decay model of the Analytic Hierarchy Process (AHP) and the Softmax function, constructing a micro-macro link model accompanied by stochastic variations. Through large-scale multi-agent simulations ($N={10}^7$) validating the model’s convergence, it was demonstrated that in long-tail distributions formed under peer pressure, macroscopic statistical distance metrics such as the Kullback-Leibler (KL) divergence ignore the fact that a small number of true signals are non-linearly suppressed, causing a statistical misinterpretation that “the error is within an acceptable range”. This implies that as long as macroscopic statistical indicators are over-trusted, signs of critical deterioration (minorities) will be structurally marginalized. Returning to the debate on “Homogeneity (Homogenität)” in German social statistics, this paper advocates that in order to realize objective “Micro-segmentation of Homogeneous Statistical Populations,” a paradigm shift from qualitative methods relying on human intuition to quantitative methods incorporating multi-criteria decision making is essential, rather than simply expanding the sample size. Full article

As global energy demand continues to grow, the inherent safety requirements for natural gas long-distance pipelines are becoming increasingly stringent. Therefore, accurately analyzing the trends in pipeline defects using multi-round internal inspection data is of great significance for enhancing pipeline inherent safety levels and reducing the risk of pipeline medium leakage. However, existing pipeline in-line inspection data alignment methods for long-distance multi-round pipeline data alignment suffer from cumbersome alignment procedures and low computational efficiency. This paper proposes an adaptive threshold dynamic time warping defect alignment method (Adaptive Dynamic Threshold-Dynamic Time Warping, ADTH-DTW) for rapidly matching multi-round in-line inspection data. A new multi-round in-line inspection data alignment framework based on valve-weld-defect is established. By integrating the DTW algorithm into each alignment stage, unnecessary manual effort is avoided, significantly improving data alignment efficiency. First, the ADTH method is used to clean redundant weld seam data in the in-line inspection data. By dynamically generating expected values and combining an intelligent point selection strategy, the method accurately identifies and removes interfering data. Additionally, valve chamber data is used to correct the overall mileage, providing a data foundation for subsequent defect alignment. Second, the dynamic time warping algorithm is used to align weld seam data and establish a data mapping table. Finally, relative displacement methods are employed to achieve defect matching. The validation results from three rounds of in-vehicle inspection data tested on-site indicate that the ADTH-DTW algorithm achieves an average 23.08% improvement in alignment accuracy compared to methods such as DTW, KL divergence, JS divergence, and linear interpolation, with computational efficiency nearly tripled. This effectively addresses the issue of incompatible computational efficiency and accuracy in existing data alignment algorithms, thereby enhancing the intrinsic safety level of natural gas long-distance pipelines. Full article

With the development of scientific machine learning (SciML), the proposal of physics-informed neural networks (PINNs) has provided a powerful paradigm for solving partial differential equations (PDEs). While PINNs perform well in solving high-dimensional PDEs, they perform worse than traditional numerical methods for low-dimensional problems. This discrepancy arose from potential convergence conflicts induced by distinct physical magnitude of loss terms. To decouple the convergence conflicts, we propose a partial derivative guided multi-branch physics-informed neural network (PDGM-PINN). Inspired by SciML, we treat both the solution and partial derivatives as dependent variables to be predicted. The partial derivatives are directly predicted by sub-branches, while the main branch approximates the PDE solution, and all branches share error backpropagation information. Furthermore, we redesign the loss function. The loss of the governing equation is computed with the solution and partial derivatives predicted by the main and sub-branches. Schwarz’s theorem and Kullback–Leibler divergence are incorporated into the loss terms as soft constraints of partial derivatives continuity and residual distributions consistency for the governing equations. We conducted comprehensive experimental evaluations on seven PDEs, and ablation experiments, sensitivity analyses, and complexity analyses were carried out to investigate the rationality of PDGM-PINN. The results demonstrate that PDGM-PINN achieves the best performance among PINN variants with the fewest trainable parameters, effectively avoiding architectural redundancy. Full article

A difficult NP-hard optimization problem, designing cryptographically robust substitution-boxes (S-boxes) necessitates a careful balancing act between several conflicting properties, such as differential uniformity and nonlinearity. Genetic Algorithms (GAs) have been widely used for this task; however, their performance is often limited by premature convergence and insufficient diversity during crossover operations. This primarily occurs because genetic algorithms commence with limited a priori knowledge. This sort of “blindness” and failure to utilize local knowledge results in diminished performance. In GA, the crossover operations facilitate the dissemination of robust candidates within the population. Conventionally, GA implements crossover for each pair of parents for diversity and a robust solution. However, this is not invariably the situation. To enhance children’s candidacy, parental diversity is quite crucial. This paper proposes a similarity-aware crossover strategy, integrated with a Siamese learning framework, to guide the genetic algorithm for improved S-box optimization with better diversity and faster convergence by utilizing parental local information. The proposed model is similarity-aware to guarantee that the GA improves parental diversity. When the parents exhibit excessive similarity, a “regressive” crossover is opted, which ensures the propagation of a parental couple with sufficient diversity to produce superior offspring. The proposed similarity-aware GA model is applied and evaluated to generate cryptographically robust and optimized S-boxes. To verify the robustness in terms of diversity, the model has been tested using three different loss functions: contrastive loss, KL divergence loss, and the suggested method of combining both loss functions to form a hybrid loss function. The effectiveness of the proposed approach is demonstrated through the generation of high-quality S-boxes with strong cryptographic properties. Full article

Knowledge distillation (KD) transfers knowledge from large language models (LLMs) to smaller or similarly sized models in order to obtain efficient yet capable systems. However, performing distillation over all tokens is computationally expensive and may weaken the transfer signal. To address this limitation, Knowledge-Filtered Distillation (KFD) is introduced as a selective distillation approach in which tokens are filtered according to the divergence $KL({\mathcal{M}}_2\mid {\mathcal{M}}_0)$ between a teacher model (${\mathcal{M}}_2$) and a base model (${\mathcal{M}}_0$), while the student model (${\mathcal{M}}_1$) is also derived from the same base model. Only tokens whose divergence exceeds a predefined threshold are distilled. For the selected tokens, the teacher distribution is normalized over the Top-5 predictions, whereas tokens outside this case receive a label-ranking bonus. The proposed conditional Top-5/bonus target design is shown theoretically to yield a lower label-focused target error than using only Top-5 normalization or only the bonus across all tokens. In addition, the KL and cross-entropy (CE) losses are balanced through a dynamically computed batch-level coefficient $\alpha$. Experiments on multiple Turkish text datasets show that KFD consistently outperforms CE-only training, achieving higher accuracy with less data and shorter training time. KFD also outperforms entropy-based token selection methods and highlights the role of student initialization in effective knowledge transfer, thereby providing an efficient and scalable distillation framework for teacher–student models of equal size. Full article

Optimizing long-term user satisfaction in sequential recommender systems is a critical challenge. Offline reinforcement learning (RL) offers a promising solution by learning recommendation policies from historical interaction logs without incurring the high costs of online exploration. However, offline RL suffers from severe distribution shift: the learned policy often overestimates the value of out-of-distribution (OOD) items, leading to unreliable recommendations and compromising user satisfaction. To address this issue, we propose a novel framework known as the Q-Learning Regularized Decision Transformer (QRDT). Built upon the Decision Transformer architecture, QRDT models recommendations as a sequence prediction task to capture complex user interest dynamics. To mitigate distribution shift, the QRDT integrates Kullback–Leibler (KL) divergence and maximum entropy regularization into the Q-value function, enabling conservative long-term value estimation while encouraging diverse exploration within the logged data distribution. Extensive experiments on four real-world Amazon e-commerce datasets (CDs, Clothing, Cellphones, and Beauty) demonstrate that the QRDT achieves competitive performance and outperforms the PGPR baseline in most scenarios. Specifically, the proposed method yields improvements of 2.99% in Hit Rate (HR), 2.19% in Normalized Discounted Cumulative Gain (NDCG), 0.94% in Recall, and 0.84% in Precision, verifying the effectiveness of our regularization approach. Full article

This study presents a novel neural parameter diffusion approach (FWA-PDiff) designed to enhance data quality. To address the limitations of conventional diffusion models—such as inefficient sampling and insufficient feature sensitivity, which may compromise output fidelity—this study introduces four key innovations. First, the proposed model introduces an adaptive recalibration of the sampling frequency in the Fourier domain to optimize feature extraction for image data. Second, a dual-channel autoencoder architecture is employed, featuring a multi-scale, fine-grained encoder (MFE) that enables the simultaneous capture of features at multiple resolutions. Third, a wavelet-attention mechanism (WA) is incorporated into the decoder to highlight subtle high-frequency details. Fourth, the proposed model introduces a hybrid loss function that combines Mean Squared Error (MSE) and Kullback–Leibler (KL) divergence to improve data reconstruction. Collectively, these improvements enable the generation of high-fidelity parameters, thereby contributing to enhanced data quality. Extensive experiments conducted on benchmark datasets—including MNIST, CIFAR-10, CIFAR-100, and STL-10—demonstrate the effectiveness of the proposed approach, which consistently achieves superior performance in improving data quality. Full article

Non-parametric information geometry has long faced an “intractability barrier”: in the infinite-dimensional setting, the Fisher–Rao metric is a weak Riemannian metric functional that lacks a bounded inverse, rendering classical optimization and estimation techniques computationally inaccessible. This paper resolves this barrier by building the statistical manifold on the Orlicz space $L_0^{\mathsf{\Phi }}\left(P_f\right)$ (the Pistone–Sempi manifold), which provides the necessary exponential integrability for score functions and a rigorous Fréchet differentiability for the Kullback–Leibler divergence. We introduce a novel Structural Decomposition of the Tangent Space ($T_{f}M=S\oplus S^{\perp }$), where the infinite-dimensional space is split into a finite-dimensional covariate subspace (S)—representing the observable system—and its orthogonal complement ($S^{\perp }$). Through this decomposition, we derive the Covariate Fisher Information Matrix (cFIM), denoted as $G_f$, which acts as the computable “Hilbertian slice” of the otherwise intractable metric functional. Key theoretical contributions include proving the Trace Theorem ($H_G\left(f\right)=\mathrm{Tr}\left(G_f\right)$) to identify G-entropy as a fundamental geometric invariant; demonstrating the Geometric Invariance of the Covariate Fisher Information Matrix (cFIM) as a covariant $(0,2)$-tensor under reparameterization; establishing the cFIM as the local Hessian of the KL-divergence; and characterizing the Efficiency Standard through a generalized Cramer–Rao Lower Bound for semi-parametric inference within the Orlicz manifold. Furthermore, we demonstrate that this framework provides a formal mathematical justification for the Manifold Hypothesis, as the structural decomposition naturally identifies the low-dimensional subspace where information is concentrated. By shifting the focus from the intractable global manifold to the tractable covariate geometry, this framework proves that statistical information is not a property of data alone, but an active geometric interaction between the environment (data), the system (covariate subspace), and the mechanism (Fisher–Rao connection). Full article

Visual saliency modeling has achieved high predictive performance in natural image domains, yet its generalization to abstract art remains limited by the lack of explicit semantic structure and the scarcity of eye-tracking data. In such semantically ambiguous contexts, understanding the underlying drivers of attention is as critical as predictive accuracy. This paper presents an interpretable, ’white-box’ saliency framework tailored to abstract art, which constructs predictions through a weighted combination of 35 modular heuristics grounded in perceptual psychology and art theory, including contrast, grouping, isolation and symmetry. Heuristic weights are optimized via a genetic algorithm and refined by a context-aware modulation mechanism that adapts to image-level visual features. Evaluation against eye-tracking data from 40 abstract paintings demonstrates that the model with the expanded activation variant produces stable, meaningful predictions while achieving a competitive KL-divergence score (1.11 ± 0.55), which is comparable to the SalGAN baseline (1.11 ± 0.53). Analysis of the optimized weights reveals strong contributions from contrast, texture, and grouping mechanisms, while nearly half of the heuristics, including most horizontal symmetry heuristics are systematically pruned by the model. Moreover, context-aware modulation reveals that these weights are not static but shift dynamically based on image-level features such as edge density and intensity variation. By prioritizing transparency over raw predictive performance, this study demonstrates that explainable saliency models can function as robust investigative tools for decoding the principles of human visual perception in data-scarce domains. Full article

Designing noise-robust parameterized quantum circuits (PQCs) is a central challenge in the noisy intermediate-scale quantum (NISQ) regime. Existing quantum architecture search methods rely on training large SuperCircuits and evaluating SubCircuits under noisy execution, resulting in high computational cost and architecture assessments that depend on task-specific optimization and device noise. In this work, we propose a training-free quantum architecture search framework based on information-theoretic expressibility measures rather than performance-based estimators. We empirically show that noise-free KL-divergence-based expressibility exhibits a consistent monotonic association with noisy task loss across diverse circuit architectures and realistic hardware noise models. Leveraging this relationship, we introduce an expressibility-guided evolutionary search that requires neither SuperCircuit training nor noisy execution during the search phase. Since expressibility is evaluated independently of hardware noise, the method is inherently device-agnostic, enabling architectures to be reused across multiple quantum devices without re-running the search. Experiments using IBM-derived Qiskit noise models demonstrate that the proposed approach achieves competitive performance compared to SuperCircuit-based baselines, while substantially reducing computational cost. These results establish expressibility as an effective information-theoretic surrogate for ranking PQC architectures under realistic noise. Full article

Advanced high-temperature fluid reactors (ARs), such as sodium fast reactors (SFRs) and molten salt cooled reactors (MSCRs) utilize high-temperature fluids at ambient pressure. To melt the fluid during reactor startup and prevent fluid freezing during cooldown, the thermal–hydraulic systems of such ARs include heater zones consisting of specific heaters with controllers, temperature sensors, and thermal insulation. The failure of heater zones due to insulation material degradation or improper installation, resulting in parasitic heat losses, can lead to fluid freezing. The detection of faults using a heat-transfer model is difficult because of a lack of knowledge of the experimental details. Data-driven machine learning of heater zone temperature time series offers a viable alternative. In this study, we benchmarked the performance of recurrent neural networks (RNNs) in an analysis of heat-up transient temperature time series of heater zones installed on a liquid sodium vessel. The RNN models include long short-term memory (LSTM) and gated recurrent unit (GRU) networks, as well as their bi-directional variants, BiLSTM and BiGRU. Anomalous temperature points were designated using a percentile-based threshold applied to residual fluctuations in the detrended temperature time series. Additionally, the impact of the exponentially weighted moving average (EWMA) method on detection accuracy was examined. The RNN models’ performance was assessed using precision, recall, and F₁ score metrics. Results demonstrated that RNN models effectively detect anomalies in temperature time series with the best models for each heater zone achieving F₁ scores of over 93%. To explain the variations in RNN model performance across different heater zones, we used Kullback–Leibler (KL) divergence to quantify the relative entropy between training and testing data, and the Detrended Fluctuation Analysis (DFA) to assess long-range temporal correlations. For datasets with strong long-range correlations and minimal relative entropy between training and testing data, GRU is the best-performing model. When the data exhibits weaker long-term correlations and a significant relative entropy between training and testing distributions, BiGRU shows the best performance. For the data sets with intermediate values of both KL divergence and DFA, the best performance is obtained with LSTM and BiLSTM, respectively. Full article

Small pest detection in ultra-high-resolution forestry images is challenging due to extreme scale variation, complex backgrounds, and limited annotated data. To address these issues, we propose SSFPDet (Semi-Supervised Forest Pest Detector), a semi-supervised object detection framework designed for low-annotation settings. Built upon the Soft Teacher paradigm, SSFPDet integrates a YOLO-T-based overlapping slicing strategy, a Top-K pseudo-label selection mechanism, and a Kullback–Leibler (KL) divergence-based distribution alignment constraint. The slicing strategy enhances small-object representation without modifying the detector backbone, while the Top-K and KL modules improve pseudo-label reliability and semantic consistency during training. Under the 20% labeled setting, SSFPDet achieves an mAP@0.5:0.95 of 46.6, outperforming the baseline by 0.7 points. Notably, small-object detection performance (AP_S) improves by 6.6 percentage points. Ablation studies confirm the complementary contributions of spatial slicing and semantic alignment. Overall, SSFPDet provides a practical and scalable solution for high-resolution forestry pest monitoring under limited supervision. Full article

To address the challenges of insufficient frequency regulation capability of individual stations, poor collaborative economic performance, and unfair benefit allocation caused by fluctuations in photovoltaic (PV) output and variations in electric vehicle (EV) connectivity during vehicle-to-grid (V2G) interactions under high-penetration PV integration, this paper proposes a coordinated optimal operation strategy for peer-to-peer (P2P) energy sharing and joint frequency regulation among multiple electric vehicle charging stations (EVCSs). First, a collaborative framework for P2P energy sharing and joint frequency regulation among EVCSs is constructed to describe the operational mechanism of inter-station energy mutual support and coordinated response to frequency regulation signals. Subsequently, an aggregate model of the dispatchable potential for EV clusters within each station is established based on Minkowski Summation (M-sum), characterizing the charging and discharging power boundaries and frequency regulation potential of the EV clusters. Meanwhile, distributionally robust chance constraints (DRCC) based on the Kullback–Leibler (KL) divergence are introduced to handle the uncertainty of PV power generation within the EVCS. On this basis, a dynamic frequency regulation output model for EV clusters and a multi-station P2P energy sharing model are designed, with the optimization objective of minimizing the total operating cost. Finally, to quantify the differential contributions of each EVCS in the collaborative operation, an asymmetric Nash bargaining benefit allocation mechanism is proposed, which incorporates a comprehensive contribution index considering both energy sharing and joint frequency regulation, The model is solved in a distributed manner using the alternating direction method of multipliers (ADMM). Simulation results demonstrate that, compared to non-cooperative operation, the frequency regulation completeness rates of the EVCSs after cooperation increase by 5.7%, 5.2%, and 4.4%, respectively; meanwhile, the total operating cost drops from CNY 16,187.61 under non-cooperative operation to CNY 15,997.47, achieving a reduction of 1.18%. The proposed strategy not only meets grid frequency regulation demands but also enhances the economic efficiency of multi-station collaborative operation and the fairness of benefit distribution. Full article

Causal inference in text data has been a demanding objective in the field of natural language processing, mainly due to the intrinsic ambiguity and context sensitivity inherent in data, inducing uncertainty. Diminishing this uncertainty is essential in identifying reliable causal connections and advancing predictive consistency. In this research, we introduce an uncertainty-aware ensemble architecture that combines multiple text embedding schemes with both linear and nonlinear classifiers to boost causal text detection. Both sparse and neural-level embeddings were employed, and then combined it with an ensemble weighting approach based on two uncertainty estimation techniques, namely entropy-based and KL divergence-based. Unlike conventional ensemble methods with uniform or fixed voting strategies, our approach assigns weights inversely proportional to classifier uncertainty, ensuring that confident models exert greater influence on the final decisions. Our results show that TF-IDF, through its effective word frequency weighting scheme, consistently outperforms other embedding techniques, achieving better performance across both linear and nonlinear classifiers on both datasets (News Corpus and CausalLM–Adjective group). The experimental results show that our uncertainty-aware ensemble approach enhances both calibration and confidence predictions. Entropy-based weighting improves confidence in the case of linear classifiers with accuracy, F1-score, entropy and prediction confidence values of 94.3%, 94.0%, 0.382 and 0.774, respectively, while in the case of nonlinear classifiers the KL divergence-based weighting acquires a better performance with an accuracy of 97.6%, F1-score of 97.2%, KL Mean value of around 0.055 and LogLoss of 0.221. Full article