Capability Of Model Research Articles

One-class anomaly detection through color-to-thermal AI for building envelope inspection

Characterizing the energy performance of building components and locating anomalies is necessary for effectively refurbishing existing buildings. It is often challenging because defects in building envelopes deteriorate without being visible. Passive infrared thermography (PIRT) is a powerful tool used in building inspection. However, thermal image interpretation requires significant domain knowledge and is prone to artifacts arising from a complex interplay of factors. As a result, PIRT-based inspections require skilled professionals, and are labor-intensive and time-consuming. Artificial intelligence (AI) holds great promise to automate building inspection, but its application remains challenging because common approaches rely on extensive labeling and supervised modeling. It is recognized that there is a need for a more applicable and flexible approach to leverage AI to assist PIRT in realistic building inspections. In this study, we present a label-free method for detecting anomalies during thermographic inspection of building envelopes. It is based on the AI-driven prediction of thermal distributions from color images. Effectively the method performs as a one-class classifier of the thermal image regions with a high mismatch between the predicted and actual thermal distributions. The algorithm can learn to identify certain features as normal or anomalous by selecting the target sample used for training. The proposed method has unsupervised modeling capabilities, greater applicability and flexibility, and can be widely implemented to assist human professionals in routine building inspections or combined with mobile platforms to automate the inspection of large areas.

Risk assessment of flammable liquid transportation on waterways: An ontology-driven dynamic Bayesian network approach

Accidents involving the transportation of flammable liquids on waterways often lead to severe consequences, highlighting the importance of effective risk assessment under conditions of uncertainty. This paper presents a novel methodology for evaluating the risk associated with the spatiotemporal evolution of flammable liquid transportation on waterways. By leveraging ontology models and dynamic Bayesian networks, the approach involves analyzing factors that impact risk, constructing a standardized knowledge representation model, and mapping this onto a dynamic Bayesian network for comprehensive risk assessment. The methodology incorporates fuzzy theory and the Best-Worst Method to calculate probabilities within the Bayesian framework, enabling detailed analysis of factor impacts on transportation risk. A practical application is illustrated through the development of a dynamic risk evaluation model for octane transportation in the Yangtze River, demonstrating the model's capability to predict risk under varying conditions. This ontology-driven Bayesian network model provides a robust foundation for making informed management decisions in the transportation of flammable liquids, effectively addressing the challenges of semantic expression and inferencing under uncertainty to enhance safety in waterway transportation.

Forecasting building operation dynamics using a Physics-Informed Spatio-Temporal Graph Neural Network (PISTGNN) ensemble

Forecasting future building operation states provides operators with comprehensive insights, allowing them to understand and optimize the factors influencing various aspects of building performance, including energy consumption. While conventional modeling tools such as EnergyPlus are widely employed to predict the behavior of buildings, they often struggle to capture the full complexity of real-world operational dynamics, as their outputs are greatly affected by the assumptions made during the modeling process and due to the stochasticity associated with real-world building operations. In this regard, this paper investigates the Physics-Informed Deep Spatio-Temporal Graph Neural Network (PISTGNN) Ensemble, which integrates residual learning and physics constraints into an encoder-decoder structured Diffusion Convolutional Recurrent Neural Network (DCRNN), to precisely estimate building operational dynamics 5 minutes in advance. The experimental results demonstrate that the Ensemble model achieved an average improvement of 44.7% in RMSE over the pure data-driven model across seasonal test sets, underscoring its robustness. Moreover, the model's predictions deviate by only 0.78% from the true values in real-world scenarios, highlighting its exceptional accuracy and reliability for practical applications. PINN integration enhances the model's capability to manage compounding errors in data-sparse regions, reducing model uncertainty.

Regional dynamic hazard assessment of rainfall–induced landslide guided by geographic similarity

Landslides triggered by rainfall are complex phenomena influenced by a multitude of condition and trigger factors. A significant challenge in the field is the accurate and interpretable assessment of large-scale landslide hazards, particularly due to the lack of consideration for the synergistic effects of multiple triggers and spatial heterogeneity. This study introduces a novel regional hazard assessment method that leverages geographic similarity to address these challenges. Our approach consists of four key steps: (1) extraction of sample information from relevant data based on the historical distribution of landslides and their influencing factors, (2) application of a scale-space algorithm to manage spatial heterogeneity, with a partition scale determined by the q-value variation, (3) optimization of sample configuration and generation criteria under the guidance of geographic similarity for enhanced spatiotemporal modeling, and (4) utilization of machine learning models to refine inductive bias and capture nonlinear relationships, enabling a quantitative estimation of hazard probabilities for each slope unit within the prediction module. We applied our P-RF + method to Yunnan Province, China, incorporating 11 condition factors and 7 trigger factors across 624 historical rainfall-induced landslides and 1248 non-landslide cases. Comparative experiments reveal that the P-RF + model substantially outperforms existing methods in accuracy and interpretability. Furthermore, a case study during the rainy season illustrates the model's capability to provide timely warning instructions for rainfall-induced landslides. These findings underscore the potential of our proposed method to offer valuable insights for disaster prevention decision-making.Graphical

A goal-oriented document-grounded dialogue based on evidence generation

Goal-oriented Document-grounded Dialogue (DGD) is used for retrieving specific domain documents, assisting users in document content retrieval, question answering, and document management. Existing methods typically employ keyword extraction and vector space models to understand the content of documents, identify the intent of questions, and generate answers based on the capabilities of generation models. However, challenges remain in semantic understanding, long text processing, and context understanding. The emergence of Large Language Models (LLMs) has brought new capabilities in context learning and step-by-step reasoning. These models, combined with Retrieval Augmented Generation(RAG) methods, have made significant breakthroughs in text comprehension, intent detection, language organization, offering exciting prospects for DGD research. However, the “hallucination” issue arising from LLMs requires complementary methods to ensure the credibility of their outputs. In this paper we propose a goal-oriented document-grounded dialogue approach based on evidence generation using LLMs. It designs and implements methods for document content retrieval & reranking, fine-tuning and inference, and evidence generation. Through experiments, the method of combining LLMs with vector space model, or with key information matching technique is used as a comparison, the accuracy of the proposed method is improved by 21.91% and 12.81%, while the comprehensiveness is increased by 10.89% and 69.83%, coherence is enhanced by 38.98% and 53.27%, and completeness is boosted by 16.13% and 36.97%, respectively, on average. Additional, ablation analysis conducted reveals that the evidence generation method also contributes significantly to the comprehensiveness and completeness.

An efficient personalized federated learning approach in heterogeneous environments: a reinforcement learning perspective

In order to address the problem of data heterogeneity, in recent years, personalized federated learning has tailored models to individual user data to enhance model performance on clients with diverse data distributions. However, the existing personalized federated learning methods do not adequately address the problem of data heterogeneity, and lack the processing of system heterogeneity. Consequently, these issues lead to diminished training efficiency and suboptimal model performance of personalized federated learning in heterogeneous environments. In response to these challenges, we propose FedPRL, a novel approach to personalized federated learning designed specifically for heterogeneous environments. Our method tackles data heterogeneity by implementing a personalized strategy centered on local data storage, enabling the accurate extraction of features tailored to the data distribution of individual clients. This personalized approach enhances the performance of federated learning models when dealing with non-IID data. To overcome system heterogeneity, we design a client selection mechanism grounded in reinforcement learning and user quality evaluation. This mechanism optimizes the selection of clients based on data quality and training time, thereby boosting the efficiency of the training process and elevating the overall performance of personalized models. Moreover, we devise a local training method that utilizes global knowledge distillation of non-target classes, which combined with traditional federated learning can effectively address the issue of catastrophic forgetting during global model updates. This approach enhances the generalization capability of the global model and further improves the performance of personalized models. Extensive experiments on both standard and real-world datasets demonstrate that FedPRL effectively resolves the challenges of data and system heterogeneity, enhancing the efficiency and model performance of personalized federated learning methods in heterogeneous environments, and outperforming state-of-the-art methods in terms of model accuracy and training efficiency.

Testing the prediction capability of trained models: sildenafil citrate solubility in propylene glycol and 1-propanol mixtures at different temperatures

ABSTRACT This study investigates the thermodynamic characteristics, equilibrium solubility, and solvation dynamics of sildenafil citrate within binary solvent systems composed of propylene glycol and 1-propanol. Utilizing both determination and computational approaches, the solubility patterns of sildenafil citrate were quantitatively assessed, revealing direct proportionality with increments in temperature and the concentration of propylene glycol. A comparative analysis using three distinct mathematical models to represent the solid–liquid phase equilibrium was conducted. Among these, the Jouyban-Acree-van’t Hoff model yielded the most comprehensive correlation by covering the effects of both temperature and solvent composition variations on the solubility, with a mean relative deviation (MRD%) of 6.4%, suggesting a robust alignment between the modelled and experimental solubility data. Previously trained models were tested to predict the solubility of the drug in mono- and mixed-solvents at various temperatures where accurate predictions were obtained with the MRD% of 19.1% (for mono-solvents), 8.1 and 14.4% (for mixed solvent systems).

Performance prediction of sintered NdFeB magnet using multi-head attention regression models

The preparation of sintered NdFeB magnets is complex, time-consuming, and costly. Data-driven machine learning methods can enhance the efficiency of material synthesis and performance optimization. Traditional machine learning models based on mathematical and statistical principles are effective for structured data and offer high interpretability. However, as the scale and dimensionality of the data increase, the computational complexity of models rises dramatically, making hyperparameter tuning more challenging. By contrast, neural network models possess strong nonlinear modeling capabilities for handling large-scale data, but their decision-making and inferential processes remain opaque. To enhance interpretability of neural network, we collected 1,200 high-quality experimental data points and developed a multi-head attention regression model by integrating an attention mechanism into the neural network. The model enables parallel data processing, accelerates both training and inference speed, and reduces reliance on feature engineering and hyperparameter tuning. The coefficients of determination for remanence and coercivity are 0.97 and 0.84, respectively. This study offers new insights into machine learning-based modeling of structure-property relationships in materials and has potential to advance the research of multimodal NdFeB magnet models.

An adaptive method for predicting bearing remaining useful life across various degradation stages

Abstract Bearing degradation is a multi-stage, multi-trend and highly complex process, significant information discrepancies and extreme imbalances exist in degradation data across different stages. These complexities hinder the accuracy of predictive model in predicting the remaining useful life throughout all stages of the bearing's degradation. In this paper, a novel prediction model based on Adaptive Convolutional Neural Network (ACNN)-Multiple Kernel Convolutional Long Short-Term Memory (MKConvLSTM) is proposed, which utilizes adaptive feature extraction and multi-scale dynamic selection to solve the problem of multi-stage, multi trend and highly complex information in bearing degradation. First, the ACNN is used to perform convolutional feature extraction and adaptive mapping on input samples, effectively distinguishing the degradation stages Then, the MKConvLSTM generates features at different time scales and dynamically selects these features to capture temporal information during the degradation process, enriching the model's capability to represent complex information and improving its predictive performance. To validate the effectiveness of the proposed model, experiments were conducted on the PHM2012 datasets and XJTU datasets. The MAE and RMSE of ACNN-MKConvLSTM reaches 0.078 and 0.099 on the first dataset, 0.086 and 0.107 on the second dataset, respectively. approximately 23% and 17% improvement Compared to the baseline model, respectively. Experimental results demonstrate that the model exhibits high accuracy and robustness in bearing remaining useful life prediction, effectively addressing the impact of feature variations across different degradation stages on prediction performance.

Enhancing person re-identification via Uncertainty Feature Fusion Method and Auto-weighted Measure Combination

Person re-identification (Re-ID) is a challenging task that involves identifying the same person across different camera views in surveillance systems. Current methods usually rely on features from single-camera views, which can be limiting when dealing with multiple cameras and challenges such as changing viewpoints and occlusions. In this paper, a new approach is introduced that enhances the capability of ReID models through the Uncertain Feature Fusion Method (UFFM) and Auto-weighted Measure Combination (AMC). UFFM generates multi-view features using features extracted independently from multiple images to mitigate view bias. However, relying only on similarity based on multi-view features is limited because these features ignore the details represented in single-view features. Therefore, we propose the AMC method to generate a more robust similarity measure by combining various measures. Our method significantly improves Rank@1 (Rank-1 accuracy) and Mean Average Precision (mAP) when evaluated on person re-identification datasets. Combined with the BoT Baseline on challenging datasets, we achieve impressive results, with a 7.9% improvement in Rank@1 and a 12.1% improvement in mAP on the MSMT17 dataset. On the Occluded-DukeMTMC dataset, our method increases Rank@1 by 22.0% and mAP by 18.4%.

Challenges with Literature-Derived Data in Machine Learning for Yield Prediction: A Case Study on Pd-Catalyzed Carbonylation Reactions.

The application of machine learning (ML) to predict reaction yields has shown remarkable accuracy when based on high-throughput computational and experimental data. However, the accuracy significantly diminishes when leveraging literature-derived data, highlighting a gap in the predictive capability of the current ML models. This study, focusing on Pd-catalyzed carbonylation reactions, reveals that even with a data set of 2512 reactions, the best-performing model reaches only an R2 of 0.51. Further investigations show that the models' effectiveness is predominantly confined to predictions within narrow subsets of data, closely related and from the same literature sources, rather than across the broader, heterogeneous data sets available in the literature. The reliance on data similarity, coupled with small sample sizes from the same sources, makes the model highly sensitive to inherent fluctuations typical of small data sets, adversely impacting stability, accuracy, and generalizability. The findings underscore the inherent limitations of current ML techniques in leveraging literature-derived data for predicting chemical reaction yields, highlighting the need for more sophisticated approaches to handle the complexity and diversity of chemical data.

Top 10+1 indicators for assessing forest ecosystem conditions: A five-decade fragmentation analysis

Globally, land use change has consistently resulted in greater losses than gains in aboveground biomass (AGB). Forest fragmentation is a primary driver of biodiversity loss and the depletion of natural capital. Measuring landscape characteristics and analyzing changes in forest landscape patterns are essential for accounting for the contributions of forest ecosystems to the economy and human well-being. This study predicts national forest distribution for 2036 and 2054 using a Cellular Automata (CA) system and assesses ecosystem conditions through landscape metrics at the patch, class, and landscape levels. We calculated 130 metrics and applied a Variance Threshold method to remove features with low variance, testing different thresholds. The first filtered-out metrics were further analysed through Principal Component Analysis combined with a Feature Importance technique to select and rank the top 10 indicators: effective mesh size, splitting index, mean radius of gyration, largest patch index, mean core area, core area percentage, Simpson's evenness index, mutual information, Simpson's diversity index, and mean contiguity index. The eleventh selected indicator is the AGB density, a structural measurement for ecosystem condition and a proxy for forest carbon storage and sequestration assessments. From 2000 to 2018, the national AGB forest carbon stock decreased from 131.5 to 91.3 Megatons (Mt) with expected values for 2036 and 2054 being 71.8 and 55.3 Mt., respectively. Landscape measurements quantitatively describe forest dynamics, providing insights into the structure, configuration, and changes characterizing landscape evolution. This research underscores the capability of CA models to map large-scale forest resources and predict future development scenarios, offering useful information for conservation and environmental management decisions. Additionally, it provides measurements to support Ecosystem Accounting by assessing forest extent and indicators of its conditions.

Finite strain continuum phenomenological model describing the shape-memory effects in multi-phase semi-crystalline networks

Thermally-driven semi-crystalline polymer networks are capable to achieve both the one-way shape-memory effect and two-way shape-memory effect under stress and stress-free conditions, therefore representing an appealing class of polymers for applications requiring autonomous reversible actuation and shape changes. In these materials, the shape-memory effects are achieved by leveraging the synergistic interaction between one or more crystalline phases and the surrounding amorphous ones that are present within the network itself. The present paper introduces a general framework for the finite strain continuum phenomenological modeling of the thermo-mechanical and shape-memory behavior of multi-phase semi-crystalline polymer networks. Model formulation, including the definition of phase and control variables, kinematic assumptions, and constitutive specifications, is introduced and thoroughly discussed. Theoretical derivations are general and easily adaptable to all cross-linked systems which include two or more crystalline domains or a single crystalline phase with a wide melting range and manifest macroscopically the one-way shape-memory effect and the two-way shape-memory effect under stress and stress-free conditions. Model capabilities are validated against experimental data for copolymer networks with two different crystalline phases characterized by well-separated melting and crystallization transitions. Results demonstrate the accuracy of the proposed model in predicting all the phenomena involved and in furnishing a useful support for future material and application design purposes.

Advanced Data Analytics and Machine Learning Driven Fraud Detection and Data Loss Prevention for Automated Incident Response in the US Healthcare Corporations

Fraud detection and data loss prevention are critical challenges facing US healthcare corporations as they strive to safeguard sensitive patient information and comply with stringent data protection regulations. The integration of advanced data analytics and machine learning has emerged as a powerful approach to enhance the efficiency and accuracy of detecting fraudulent activities and preventing data breaches. This study explores the application of machine learning-driven solutions in automating incident response for healthcare data security. The research begins by examining the current landscape of data analytics and machine learning in fraud detection, emphasizing the limitations of traditional methods. Through an extensive literature review and analysis of case studies within the US healthcare industry, the paper identifies key areas where advanced technologies can bridge existing gaps. The methods section outlines the data collection process, the machine learning algorithms implemented, and the evaluation metrics used to measure model performance. Results demonstrate the enhanced detection accuracy and prompt response capabilities of machine learning models compared to conventional techniques. The discussion delves into the implications of these findings, showcasing the transformative potential of automated incident response systems in reducing response times and mitigating data loss risks. Although promising, the study acknowledges limitations in data variability and model generalizability, suggesting avenues for further research. The paper concludes with strategic recommendations for adopting machine learning solutions in healthcare security protocols. By highlighting best practices and policy recommendations, this research aims to provide a roadmap for healthcare corporations seeking to strengthen their data protection frameworks. The insights presented underscore the pivotal role of advanced data analytics and machine learning in fortifying healthcare data security against evolving cyber threats.

GarVerseLOD: High-Fidelity 3D Garment Reconstruction from a Single In-the-Wild Image using a Dataset with Levels of Details

Neural implicit functions have brought impressive advances to the state-of-the-art of clothed human digitization from multiple or even single images. However, despite the progress, current arts still have difficulty generalizing to unseen images with complex cloth deformation and body poses. In this work, we present GarVerseLOD, a new dataset and framework that paves the way to achieving unprecedented robustness in high-fidelity 3D garment reconstruction from a single unconstrained image. Inspired by the recent success of large generative models, we believe that one key to addressing the generalization challenge lies in the quantity and quality of 3D garment data. Towards this end, GarVerseLOD collects 6,000 high-quality cloth models with fine-grained geometry details manually created by professional artists. In addition to the scale of training data, we observe that having disentangled granularities of geometry can play an important role in boosting the generalization capability and inference accuracy of the learned model. We hence craft GarVerseLOD as a hierarchical dataset with levels of details (LOD) , spanning from detail-free stylized shape to pose-blended garment with pixel-aligned details. This allows us to make this highly under-constrained problem tractable by factorizing the inference into easier tasks, each narrowed down with smaller searching space. To ensure GarVerseLOD can generalize well to in-the-wild images, we propose a novel labeling paradigm based on conditional diffusion models to generate extensive paired images for each garment model with high photorealism. We evaluate our method on a massive amount of in-the-wild images. Experimental results demonstrate that GarVerseLOD can generate standalone garment pieces with significantly better quality than prior approaches while being robust against a large variation of pose, illumination, occlusion, and deformation. Code and dataset are available at garverselod.github.io.

Machine learning-informed liquid-liquid phase separation for personalized breast cancer treatment assessment.

Breast cancer, characterized by its heterogeneity, is a leading cause of mortality among women. The study aims to develop a Machine Learning-Derived Liquid-Liquid Phase Separation (MDLS) model to enhance the prognostic accuracy and personalized treatment strategies for breast cancer patients. The study employed ten machine learning algorithms to construct 108 algorithm combinations for the MDLS model. The robustness of the model was evaluated using multi-omics and single-cell data across 14 breast cancer cohorts, involving 9,723 patients. Genetic mutation, copy number alterations, and single-cell RNA sequencing were analyzed to understand the molecular mechanisms and predictive capabilities of the MDLS model. Immunotherapy targets were predicted by evaluating immune cell infiltration and immune checkpoint expression. Chemotherapy targets were identified through correlation analysis and drug responsiveness prediction. The MDLS model demonstrated superior prognostic power, with a mean C-index of 0.649, outperforming 69 published signatures across ten cohorts. High-MDLS patients exhibited higher tumor mutation burden and distinct genomic alterations, including significant gene amplifications and deletions. Single-cell analysis revealed higher MDLS activity in tumor-aneuploid cells and identified key regulatory factors involved in MDLS progression. Cell-cell communication analysis indicated stronger interactions in high-MDLS groups, and immunotherapy response evaluation showed better outcomes for low-MDLS patients. The MDLS model offers a robust and precise tool for predicting breast cancer prognosis and tailoring personalized treatment strategies. Its integration of multi-omics and machine learning highlights its potential clinical applications, particularly in improving the effectiveness of immunotherapy and identifying therapeutic targets for high-MDLS patients.

Shaping Vectors

This article investigates how the word embeddings at the heart of large language models are shaped into acceptable meanings. We show how such shaping follows two educational logics. The use of benchmarks to discover the capabilities of large language models exhibit similar features to Foucault’s disciplining school enclosures, while the process of reinforcement learning is framed as a modulation made explicit in Deleuze’s control societies. The consequences of this shaping into acceptable meaning is argued to result in semantic subspaces. These semantic subspaces are presented as the restricted lexical possibilities of human-machine dialogic interaction, and their consequences are discussed.

MV2MV: Multi-View Image Translation via View-Consistent Diffusion Models

Image translation has various applications in computer graphics and computer vision, aiming to transfer images from one domain to another. Thanks to the excellent generation capability of diffusion models, recent single-view image translation methods achieve realistic results. However, directly applying diffusion models for multi-view image translation remains challenging for two major obstacles: the need for paired training data and the limited view consistency. To overcome the obstacles, we present a first unified multi-view image to multi-view image translation framework based on diffusion models, called MV2MV. Firstly, we propose a novel self-supervised training strategy that exploits the success of off-the-shelf single-view image translators and the 3D Gaussian Splatting (3DGS) technique to generate pseudo ground truths as supervisory signals, leading to enhanced consistency and fine details. Additionally, we propose a latent multi-view consistency block, which utilizes the latent-3DGS as the underlying 3D representation to facilitate information exchange across multi-view images and inject 3D prior into the diffusion model to enforce consistency. Finally, our approach simultaneously optimizes the diffusion model and 3DGS to achieve a better trade-off between consistency and realism. Extensive experiments across various translation tasks demonstrate that MV2MV outperforms task-specific specialists in both quantitative and qualitative.

MULTICLASS CLASSIFICATION FOR STUNTING PREDICTION USING DEEP NEURAL NETWORKS

Stunting is a chronic nutritional issue that hinders child growth and leads to serious long-term health and developmental impacts, particularly in developing countries. Therefore, early and accurate prediction of stunting is crucial for implementing effective interventions. This research aims to develop a multiclass classification model based on Deep Neural Networks (DNNs) to predict stunting status. The model is trained using a comprehensive dataset that encompasses various health variables related to stunting. The research process includes data collection, data preprocessing, dataset splitting, and training and evaluation of the DNNs model. The model can classify stunting status into four categories: stunted, severely stunted, normal, and tall. Further analysis is conducted to evaluate the influence of various parameters on the model's performance, including dataset splitting ratios (80:20 and 70:30) and learning rates (0.001, 0.0001, and 0.00001). The results show that a learning rate of 0.0001 yields the highest prediction accuracy, at 93.64% and 93.83% for the two data-splitting schemes. This indicates that this learning rate has achieved an optimal balance between convergence speed and the model's generalization capability. Additionally, the developed DNNs model can identify complex patterns hidden within the data without being affected by noise. These findings confirm that appropriate parameter selection, particularly the dataset splitting ratio and learning rate, can significantly enhance the DNNs model's ability to identify complex data patterns.

Mixed style network based: A novel rotating machinery fault diagnosis method through batch spectral penalization

The unsupervised fault diagnosis of rotating machinery holds significant importance, but it still faces numerous complex challenges. For instance, traditional convolutional neural networks often overlook inter-channel relationships, resulting in poor generalization and requiring manual adjustment of architecture parameters for different tasks. Additionally, traditional domain adversarial transfer learning has insufficient research on feature discriminability, leading to less distinguishable features. To address these issues, this paper proposes a MixStyle network based on the SE attention mechanism. This method achieves dynamic weight allocation through the SE attention mechanism, which is simple in design and introduces few additional parameters. By employing the MixStyle method for probabilistic mixed-domain training, the diversity of the source domain is increased, thereby improving the model's generalization capability. Since the principal singular vector enhances feature transferability, this paper penalizes the largest singular value through Batch Spectral Penalization to enhance other feature vectors, improving feature discriminability and domain adversarial performance. Experimental results show that the proposed method demonstrates outstanding performance in the task of unsupervised fault diagnosis for rotating machinery.