Search Results (48)

Variations in mechanical load constitute a dominant source of domain shift in data-driven fault diagnosis of rotating machinery, causing models trained under one operating condition to degrade sharply when deployed under another. This work addresses the problem through unsupervised domain adaptation (UDA)—which transfers diagnostic knowledge from a labeled source condition to an unlabeled target condition by aligning their feature distributions—and introduces Padé Approximant Neural Networks (PadéNets) as compact yet highly expressive feature extractors. One-dimensional PadéNet encoders are embedded into three established adaptation frameworks—Deep CORAL, Domain-Adversarial Neural Networks (DANNs), and Conditional Domain-Adversarial Networks (CDANs)—to learn load-invariant representations without any labeled target data. On the Case Western Reserve University benchmark, where the models operate directly on raw time-domain vibration signals, replacing conventional convolutional encoders with PadéNets consistently improves cross-load diagnostic accuracy, reaching up to 99.28% average target-domain accuracy at a low parameter count. To assess generalization to a more demanding setting, the CDAN–PadéNet configuration is further evaluated on frequency-domain representations of the Paderborn University dataset, where domain shift arises from simultaneous variation of load torque and radial force on bearings with real accelerated-lifetime damage, attaining 99.84% average accuracy across six cross-condition transfer tasks while requiring fewer parameters than competing methods. These results establish PadéNet-enhanced UDA as an accurate, broadly applicable approach for robust bearing fault diagnosis under varying operating conditions, with a reduced parameter count suited to resource-constrained embedded platforms. Full article

Objectives: To assess the feasibility and reproducibility of predicting antimicrobial resistance (AMR) in Escherichia coli from MALDI-TOF mass spectrometry data using a standardized, open-source machine learning (ML) workflow, we systematically compared four ML algorithms, evaluated the impact of culture conditions, extract storage, and spectral preprocessing on model performance, and validated results through nested cross-validation with statistical significance testing. Methods: A total of 282 clinical E. coli isolates were analyzed. Two MALDI-TOF MS datasets were generated from freshly cultured extracts (T₁) and recultured isolates one year later (T₃), yielding 4468 spectra. A third dataset from the T₁ extracts stored at −20 °C for one year (T₂) was evaluated for spectral stability but excluded from primary modeling likely due to storage-induced degradation. Protein spectra (m/z 2000–15,000) were preprocessed using an in-house developed MALDI-TOF preprocessing pipeline (MTPP) comprising variance stabilization, Savitzky–Golay smoothing, SNIP baseline correction, TIC normalization, LOWESS alignment, and MAD-based peak detection (SNR ≥ 3), yielding 121 m/z features. Four classifiers—Random Forest (RF), Logistic Regression, Support Vector Machine, and Gradient Boosting—were trained to predict resistance to 11 antibiotics using nested cross-validation: outer GroupShuffleSplit (5-fold, isolate-level) for evaluation and inner GroupKFold for recursive feature elimination (RFECV) and hyperparameter tuning (RandomizedSearchCV). Classification thresholds were optimized via the precision–recall curve. Model performance was assessed using AUROC, AUPRC, F1-score, Matthews Correlation Coefficient (MCC), and bootstrap 95% confidence intervals (1000 replicates). Pairwise model comparisons were tested with McNemar’s chi-squared test. Results: Among the 12 antibiotics included in the analysis (meropenem excluded for absence of resistance), resistance prevalence ranged from 1.1% (colistin) to 59.9% (amoxicillin). Colistin was subsequently also excluded from ML modeling due to insufficient resistant isolates (n = 3), leaving 11 antibiotics for prediction. The best predictive performance was observed for ciprofloxacin (AUROC 0.76 [95% CI 0.74–0.77]; F1 0.54; MCC 0.38) and ceftazidime (AUROC 0.68 [0.65–0.71]; F1 0.36; MCC 0.29), using 13 and 37 RFECV-selected features, respectively. Amoxicillin achieved the highest F1-score (0.76), driven by high recall (0.98) but modest AUROC (0.58). No meaningful predictive signal was detected for amikacin, cefepime, or tigecycline (AUROC ≤ 0.57, F1 ≤ 0.17), attributable to extreme class imbalance, and no robust multi-peak resistance signature was detected in this dataset. McNemar’s test confirmed that RF significantly outperformed Logistic Regression for all antibiotics (p < 0.01), while Gradient Boosting performed comparably to RF for ciprofloxacin (p = 0.17) and ceftazidime (p = 0.28). Frozen extracts (T₂) produced lower spectral similarity and were excluded from model training; the aligned T₁₊₃ dataset yielded the most stable performance across metrics. Conclusions: Machine learning analysis of MALDI-TOF spectra enables reproducible AMR prediction for selected antibiotics in E. coli, with ciprofloxacin and ceftazidime showing the strongest signal. Nested isolate-level cross-validation, multi-model comparison with statistical testing, and open-source code provide a transparent, reproducible foundation for integrating ML-assisted MALDI-TOF analysis into diagnostic AMR surveillance. Extract storage at −20 °C degrades spectral quality and should be avoided in ML training workflows. Full article

With the rapid development of intelligent manufacturing, edge computing, and industrial and financial–industrial digital systems, large volumes of multisource hardware sensing signals are continuously generated in complex production environments, including environmental, electrical, vibration, network communication, and device operational signals. Owing to the heterogeneity, asynchrony, noise interference, and disturbance sensitivity of these signals, conventional state prediction methods often fail to sufficiently characterize the dynamic response relationships among different sensing sources and cannot maintain stable prediction performance under non-stationary scenarios such as load surges, network congestion, and device anomalies. To address these challenges, a multisource hardware sensing signal fusion network is proposed for the edge-computing and digital production test scenario of an intelligent equipment manufacturing enterprise in Hebei Province, China, with the aim of achieving robust state prediction and anomaly perception in complex digital systems. In the proposed method, environmental sensing, device power, edge-node operation, vibration monitoring, network communication, and system output states are uniformly modeled as multisource engineering sensing signals, and an end-to-end prediction framework is constructed with cross-source sensing signal alignment to facilitate temporal coherence, disturbance-aware residual correction to substantially mitigate disturbance contamination, and context-adaptive fusion. Experimental results show that the proposed method achieves the best performance in the overall state prediction task, with MAE, RMSE, MAPE, and $R^2$ reaching $0.0968$, $0.1457$, $8.12$%, and $0.9416$, respectively, outperforming baseline methods including ARIMA, XGBoost, LightGBM, LSTM, TCN, Transformer, Attention Fusion, and Multimodal Transformer. In the disturbance robustness experiment, the Event-MAE and Event-RMSE of the proposed method are reduced to $0.1126$ and $0.1694$, respectively, with an Avg. Drop of only $28.98$%, indicating that more stable responses can be achieved under non-stationary disturbance scenarios. In the abnormal-state recognition task, Accuracy, Precision, Recall, and F1-score values of $94.32$%, $93.76$%, $92.85$%, and $93.30$% are achieved, respectively. The results demonstrate that the proposed method can effectively improve the state prediction accuracy, disturbance robustness, and anomaly warning capability of multisource hardware sensing data in complex industrial and financial–industrial digital systems, thereby providing an effective modeling scheme for intelligent monitoring and engineering decision-making in AI-driven industrial and financial sensing scenarios. Full article

Medical image analysis is a cornerstone of modern healthcare, yet conventional single-modal deep learning often struggles with the unique physical constraints and structural variability inherent in data acquired from diverse medical sensors. Recently, Vision-Language Models (VLMs) have sparked a paradigm shift by bridging the semantic gap between visual sensor signals and clinical narratives. Following the PRISMA guidelines, 167 representative studies are systematically synthesized in this review to provide a comprehensive roadmap of VLM technological evolution and clinical utility. First, rather than treating VLMs as generic feature extractors, their underlying mechanisms are uniquely distilled into seven core operational principles, which are then explicitly mapped to downstream applications such as few-shot diagnosis, prompt-driven segmentation, and multi-task foundation models. To facilitate intuitive evaluation, a rigorous quantitative cross-comparison of current benchmark architectures is presented. Crucially, this review goes beyond highlighting successes by critically assessing prevalent clinical bottlenecks, including zero-shot segmentation failures, multi-modal hallucinations in diagnosing rare diseases, and the prohibitive computational complexity associated with 3D volumes and gigapixel whole slide images. Finally, a novel, forward-looking framework is proposed: the transition from static “image-text alignment” to dynamic “multi-source sensor-driven intelligence”. By addressing both physical sensor constraints and algorithmic limitations, this survey offers actionable insights for developing trustworthy, sensor-aware clinical diagnostic agents. Full article

Multi-source heterogeneous data in discrete manufacturing shop floors, including vibration signals, equipment logs, visual monitoring data, and handwritten production reports, exhibit significant differences in modality and semantic representation. Traditional fusion methods often fail to bridge the semantic gap between low-level sensing signals and high-level manufacturing cognition, limiting intelligent anomaly analysis and decision-making capability. To address this issue, this paper proposes a semantic translation and fusion framework for industrial heterogeneous data based on Knowledge Graph (KG), Retrieval-Augmented Generation (RAG), and Large Language Models (LLMs). First, a unified semantic translation mechanism is developed to convert multimodal industrial data into structured semantic representations for cross-modal alignment. Second, an industrial knowledge graph and RAG mechanism are introduced to integrate process knowledge, maintenance manuals, and historical fault records into the reasoning process. Third, an LLM-driven reasoning framework is designed for multimodal semantic fusion, anomaly identification, causal analysis, and optimization recommendation generation. In addition, a digital twin-based visualization interface is constructed to realize real-time interaction between production lines, industrial data, and intelligent cognitive reports. Experimental results demonstrate that the proposed framework significantly improves industrial reasoning accuracy, anomaly analysis correctness, and response efficiency compared with general-purpose LLMs, providing an effective solution for intelligent cognition and decision-making in discrete manufacturing systems. Full article

To address the challenges of strong signal noise, pronounced cross-modal asynchrony, high subjectivity in manually defined state labels, and insufficient model stability under extreme abnormal conditions in multi-source sensor systems, a low-noise system state prediction and anomaly detection method based on multimodal sensor signals and self-supervised representation learning is proposed. Environmental sensing data, device status data, network transmission data, operational behavior data, and event log data are uniformly modeled as system state perception signals. A temporal masking-based state structure modeling method, a state-oriented contrastive learning representation constraint mechanism, and a state representation and downstream prediction task alignment strategy are designed to learn stable, transferable, and interpretable system state features. Experimental results demonstrate that the proposed method achieves the best performance in multimodal sensor state prediction and anomaly detection tasks, with mean squared error (MSE), mean absolute error (MAE), and root mean square error (RMSE) values of $0.0167$, $0.0856$, and $0.1291$, respectively, outperforming baseline models such as GARCH, MLP, LSTM, TCN, and Transformer. Meanwhile, $IC$, $RankIC$, and $AUC$ reach $0.494$, $0.460$, and $0.815$, respectively, indicating stronger state-ranking capability and improved discrimination between high-abnormality and low-abnormality states. At the classification recognition level, superior accuracy, precision, recall, and $F1$-score are also achieved by the proposed method, suggesting that potential abnormal states can be identified more accurately. Ablation experiments verify the effectiveness of multimodal fusion, temporal masking modeling, self-supervised contrastive constraints, and task alignment strategies. Robustness experiments further show that lower prediction errors and higher $AUC$ can still be maintained under high-fluctuation and extreme-shock states, demonstrating strong noise resistance, stability, and practical application potential in complex sensor system scenarios. Full article

The use of biodiesel blends in heavy-duty diesel engines changes the relationship between engine operating conditions, fuel properties, and exhaust emissions, which may limit the reliability of data-driven emission models trained under a single fuel condition. This study investigates the cross-fuel transferability of virtual emission sensors for a heavy-duty diesel engine operating on B7 and B20 fuel blends. The analysis was carried out for three target signals: nitrogen oxides concentration, hydrocarbon concentration, and dry carbon dioxide concentration, using data from the World Harmonized Transient Cycle (WHTC) and World Harmonized Stationary Cycle (WHSC) tests. A structured modelling workflow was developed, including signal time alignment, construction of baseline, dynamic, and memory-based features, feature selection, and separate evaluation scenarios: within-domain, cross-cycle, and cross-fuel transfer. Three tree-based regression algorithms were compared: Random Forest (RF), Histogram-Based Gradient Boosting (HGB), and Extreme Gradient Boosting (XGBoost). XGBoost achieved the best predictive performance in the source domain and was selected as the reference model. The results showed that a change in cycle characteristics led to a significant decrease in predictive performance, whereas the transition from B7/WHTC to B20/WHTC resulted in a clearly smaller drop in the evaluation metrics. The relationship between engine operating signals and emission response remained partially transferable across fuels. The highest stability was observed for carbon dioxide, intermediate stability for nitrogen oxides, and the lowest stability for hydrocarbons. The findings support the development of robust data-driven virtual sensing methods for emission monitoring and calibration of heavy-duty diesel engines operating with biodiesel blends. Full article

Cross-domain emotion recognition based on electroencephalogram (EEG) is a challenging task, as EEG signals collected from different subjects or at different moments exhibit significant differences in distribution. How to enable deep learning model to learn the common feature space and reduce the distribution differences between the source and target domains is an important research direction. For this problem, we propose a Supervised Contrastive Variational AutoEncoder Network (SCVAE-Net), which possesses enhanced abilities for extracting consistent features across source and target domains, thereby improving cross-domain EEG emotion recognition performance. Specifically, this method utilizes the reconstruction mechanism and latent space probabilization of VAE to obtain intermediate features that are more consistent and transferable. Furthermore, the maximum mean discrepancy loss is employed to further reduce the distribution discrepancy of these features. To alleviate the degradation of discriminative ability during domain alignment, we introduce multi-view supervised contrastive learning in multi-source domains to enhance the intra-class consistency and inter-class separability of latent features. Under the cross-subject and cross-session settings, SCVAE-Net achieves accuracies of 95.01%/96.84% on SEED and 74.94%/79.44% on SEED-IV, respectively. These experimental results demonstrate the effectiveness of the proposed method in cross-domain EEG emotion recognition. Full article

Early-exit cascades over a frozen convolutional backbone enable adaptive inference but suffer from three sources of train–inference mismatch: branches train on samples they will never see at inference; their per-class precision thresholds are calibrated on the wrong distribution; the standard cross-entropy target on backbone argmax labels discards the backbone’s uncertainty signal. We close all three gaps with CalexNet (cascade-aligned early exits), a training-recipe-only modification: branches train under continuously weighted importance sampling that matches the cascade-survivor distribution; per-class precision thresholds are calibrated on the actual cascade-survivor subset of the validation set; the classification head is trained against the backbone’s full softmax via a temperature-scaled KL objective. Combined with an augmented prototype-pooling branch head, CalexNet is evaluated on ResNet18 and ResNet50 backbones across CIFAR-100 (20-supe-class coarse, the harder primary setting) and CINIC-10 (10-class, the easier cross-validation counterpart). On the accuracy–FLOPs Pareto frontier, CalexNet matches or exceeds three published baselines (PTEEnet, ZTW, BoostNet) and a within-paper “no-alignment, no-KD” reference. The largest gains appear in the practically relevant 30–70% FLOPs-reduction regime and show consistent trends across $n=3$ training seeds. CalexNet requires no inference-time architectural change and is a drop-in for any frozen-backbone early-exit cascade. Full article

Intelligent detection in modern manufacturing, healthcare, process industries, and structural monitoring is fundamentally enabled by heterogeneous sensor systems. Rather than being viewed as a purely image-centered recognition task, intelligent detection is more appropriately formulated as a sensor-driven state inference problem in which sensing physics, signal quality, temporal synchronization, modality availability, and deployment conditions jointly determine what can be reliably detected, localized, interpreted, and acted upon. Against this background, this review provides a structured synthesis of the field through three coupled dimensions, namely methods, systems, and governance, and organizes the literature around four recurring engineering components: signal unification, representation unification, alignment mechanisms, and robustness mechanisms. Using a structured review protocol with explicit source selection, screening, and study coding, the paper traces the methodological evolution from traditional feature-engineering and model-based pipelines to deep learning for visual, temporal, multimodal, generative, and mechanism-constrained sensing, and further to foundation-model-based and multimodal sensor intelligence. Cross-domain evidence is synthesized from industrial defect detection, fault diagnosis, remaining useful life prediction, non-destructive testing, structural health monitoring, medical lesion analysis, and process monitoring. The review argues that recent progress has substantially strengthened learned representations, multimodal interaction, and semantic extensibility, but has not removed persistent constraints arising from domain shift, missing modalities, calibration instability, privacy-preserving collaboration, and edge-side resource limits. Accordingly, the central challenge is no longer how to optimize isolated detection models, but how to build sensor-enabled intelligent systems that remain physically grounded, trustworthy, transferable, and maintainable under real operational conditions. On this basis, the paper concludes by identifying future directions in mechanism-aware modeling, trustworthy evaluation, missing-modality-robust multimodal systems, privacy-preserving cross-site collaboration, and edge-native lifecycle-aware deployment. Full article

Background/Objectives: Adversarial domain adaptation methods are widely used in EEG-based emotion recognition to reduce the influence of individual differences and the non-stationary characteristics of electroencephalogram (EEG) signals. Most existing methods employ binary domain discriminators to align source and target domains at the global distribution level. However, such strategies often neglect the potential multimodal structure of emotional EEG data and the asymmetric emotional processing characteristics of the left and right hemispheres. To address these issues, this study proposes a Bi-Hemispheric Adversarial Domain Adaptation Neural Network (BiHADA) for EEG-based emotion recognition. Methods: In the proposed BiHADA framework, the conventional binary domain discriminator is extended into a multimodal discriminator by incorporating the label structure information of source-domain data into the domain discrimination process. This design encourages features belonging to the same emotional category to be aligned across domains and promotes positive knowledge transfer. In addition, dual adversarial domain adaptation branches are constructed to model the left and right hemispheres separately, enabling the network to capture hemisphere-specific emotional representations. Furthermore, discriminator-derived perplexity is introduced to evaluate the distribution alignment quality of target samples and to adaptively determine the weights of the corresponding hemisphere classifiers, thereby reducing the influence of poorly aligned samples during the final decision stage. Results: Experiments on the SEED dataset show that BiHADA achieves classification accuracies of 86.82% and 92.71% in cross-subject and cross-session tasks, respectively. These results demonstrate that the proposed method can effectively improve the transferability and discriminability of EEG emotional features under different domain adaptation scenarios. Conclusions: The proposed BiHADA method enhances EEG-based emotion recognition by jointly considering class-structure-guided domain alignment, hemispheric functional asymmetry, and branch-wise adaptation quality. The results suggest that incorporating source-domain label structure and hemisphere-specific adaptation can improve cross-domain EEG emotion recognition performance. Full article

Plastic recycling technologies are rapidly being reoriented toward process, operations, and quality-centered innovation, driven by the circular economy and digital transformation. Using 64,639 triadic patents (2005–2024), this study applies International Patent Classification (IPC) portfolio, co-occurrence network, and BERTopic analyses to compare technological structures before and after 2015. Since 2015, data- and AI-enabled sorting and process optimization (IPC class G06), tracking and connectivity (IPC class H04), collection and logistics (IPC class B65), water treatment (IPC class C02), and quality modification/compounding (IPC class C09) have expanded, while organic chemistry (IPC class C07), signal-processing circuitry (IPC class H03), and petroleum/fuel conversion (IPC class C10) have declined. G06 and H04 together account for approximately 29% of the total portfolio and record the largest share increases (+1.63 and +1.28 percentage points); water treatment (C02F) and quality correction (C09K) expand by 0.62 and 0.38 percentage points, while organic chemistry (C07) shows the largest decline (−2.16 percentage points). Topic modeling identifies 10 topics in 2005–2014 and 11 in 2015–2024, with the later period newly featuring reverse logistics for reusable packaging, remanufacturing, chemical recycling for packaging, and data sources. Cross-domain network linkages rise from 49 to 68, with processing–logistics and post-treatment–standardization combinations showing the strongest structural strengthening. Industrially, these findings offer reference signals for firms aligning R&D and IP portfolios with domains of concentrated innovation, particularly AI-enabled sorting, digital connectivity, and feedstock quality correction. For policy, the strengthening of cross-domain linkages suggests that support for sorting infrastructure, traceability and data standards, and quality certification frameworks targets where R&D effort is most concentrated. Full article

With increasing demand for fine-scale ecological management under carbon neutrality frameworks, multi-temporal assessment of carbon stock change (ΔC) at the individual-plant scale has become essential for understanding plant-level carbon dynamics and supporting management decisions. However, methodologies for repeated monitoring at this scale remain fragmented, showing limited cross-temporal comparability, weak cross-scale consistency, and insufficient integration across methods. Existing approaches can be grouped into three pathways: (i) process-based methods derived from CO₂ exchange measurements, (ii) state-based approaches estimating biomass and ΔC, and (iii) sensing-based approaches using structural, spectral, thermal, and fluorescence signals. These approaches offer complementary strengths, yet none simultaneously achieve high accuracy, temporal continuity, and operational scalability for multi-temporal ΔC estimation. Among these, stock-based and structural approaches form the primary estimation pathways, while flux-based and functional sensing methods provide complementary constraints. This review synthesizes and compares these approaches in terms of their theoretical basis, spatial support, temporal characteristics, and uncertainty structures. To address the lack of methodological integration, we propose a structure–function–scale framework that links heterogeneous observations across spatial and temporal domains and emphasizes cross-scale consistency as a prerequisite for reliable ΔC estimation. Within this framework, we further examine how multi-source integration can connect structural and functional observations through segmentation, co-registration, scaling, temporal alignment, and uncertainty propagation. By integrating traditional measurement logic with emerging remote sensing technologies, this review provides a unified methodological framework for ΔC estimation and identifies key directions for advancing fine-scale carbon monitoring, spatiotemporally consistent data fusion, uncertainty-aware inference, and MRV-oriented verification systems. Full article

Rapid and reliable tea classification is valuable for routine product screening, yet conventional sensory or physicochemical methods are subjective or time-consuming. Electronic nose (E-nose) sensing provides a fast alternative, but performance often degrades under domain shifts caused by different tea types, commercial categories, or acquisition conditions. This study proposes MGDA-Net, a multi-granularity domain adversarial network for cross-domain tea classification using E-nose time-series signals. MGDA-Net learns local temporal dynamics via a CNN branch and global contextual dependencies via a self-attention branch, and fuses them through an adaptive gating module. A branch-level adversarial alignment strategy is introduced to reduce source–target discrepancy at both local and global feature levels. A three-stage training procedure, consisting of source pretraining, adversarial alignment, and target fine-tuning, enables knowledge transfer from a labeled green tea source-domain to two target tasks. Experiments on oolong tea commercial-category classification (6 classes) and jasmine tea retail price-level classification (8 classes) show that MGDA-Net achieves mean accuracies of 99.31 ± 0.69% and 99.38 ± 0.51% over 10 independent runs, substantially outperforming all compared baseline methods. Ablation studies, feature-space analyses, and label-efficiency experiments further confirm the contribution of each component and show that MGDA-Net maintains mean accuracies above 87% when only 40% of the target-domain labels are used for fine-tuning. These findings suggest that MGDA-Net is a promising approach for cross-domain tea classification using E-nose data. Full article

Stock price sequences are characterized by pronounced nonlinearity, non-stationarity, and multi-scale volatility. They are further influenced by complex, multi-source factors, such as macroeconomic conditions and market behavior, making high-precision forecasting highly challenging. Existing approaches are limited by noise and multi-dimensional market features, as well as difficulties in balancing prediction accuracy with model complexity. To address these challenges, we propose Wavelet Entropy and Cross-Attention Network (WECA-Net), which combines wavelet decomposition with a multimodal cross-attention mechanism. From an information-theoretic perspective, stock price dynamics reflect the time-varying uncertainty and informational complexity of the market. We employ wavelet entropy to quantify the dispersion and uncertainty of energy distribution across frequency bands, and multiscale entropy to measure the scale-dependent complexity and regularity of the time series. These entropy-derived descriptors provide an interpretable prior of “information content” for cross-modal attention fusion, thereby improving robustness and generalization under non-stationary market conditions. Experiments on Chinese stock indices, A-Share, and CSI 300 component stock datasets demonstrate that WECA-Net consistently outperforms mainstream models in Mean Absolute Error (MAE) and R² across all datasets. Notably, on the CSI 300 dataset, WECA-Net achieves an R² of 0.9895, underscoring its strong predictive accuracy and practical applicability. This framework is also well aligned with sensor data fusion and intelligent perception paradigms, offering a robust solution for financial signal processing and real-time market state awareness. Full article