Feature Learning Method Research Articles

Image Manipulation Detection Based on Irrelevant Information Suppression and Critical Information Enhancement

The image manipulation detection localization task differs from traditional computer vision tasks in that we focus more on capturing subtle and generic manipulation detection features in images. In this paper, we propose a novel method called irrelevant visual information suppression, which aims to alleviate the interference of irrelevant visual information in images on manipulation detection feature extraction, thereby obtaining generic manipulation traces that are more subtle and unrelated to semantic visual information. In general, most manipulation operations leave traces at manipulation edges. Therefore, we introduce a specially designed manipulated edge information enhancement branch aimed at identifying these edge artifacts more accurately. We construct a dual-branch network, where each branch uses ResNet-50 as the backbone to capture as many multi-scale manipulation features as possible. Finally, we adopt a multi-view feature learning method that combines the manipulated edge information enhancement branch with the irrelevant visual information suppression branch and is trained with multi-scale (pixel/edge/image/irrelevant visual information suppression) supervision. To validate the effectiveness of the proposed method, we conducted extensive experiments using five image manipulation localization datasets, including CASIAv1, CASIAv2, COVER, Columbia, and NIST16. The experimental results demonstrate that our proposed method can outperform state-of-the-art methods by a significant margin in terms of F1 score. Taking CASIAv1, COVER, and Columbia datasets as examples, compared with MVSS-Net published in ICCV 2021, our method has improved F1 scores by 7.1%, 6.3%, and 12.5%, respectively. The code used in this paper can be found at the following URL: https://github.com/ginwins/ISIE-Net .

Discovering patient groups in sequential electronic healthcare data using unsupervised representation learning

IntroductionUnsupervised feature learning methods inspired by natural language processing (NLP) models are capable of constructing patient-specific features from longitudinal Electronic Health Records (EHR).DesignWe applied document embedding algorithms to real-world paediatric intensive care (PICU) EHR data to extract patient-specific features from 1853 patients’ PICU journeys using 647 unique lab tests and medication events. We evaluated the clinical utility of the patient features via a K-means clustering analysis.ResultsWe trained a document embedding model under a unique evaluation pipeline and obtained latent patient feature vectors for all 1853 patients. We performed unsupervised clustering to the patient vectors as a downstream analysis and obtained 5 distinct clusters via hyperparameter optimisation. Significant variations (p<0.0001) within both patient characteristics and surgery intervention and diagnostic profiles were detected.ConclusionThe K-means clustering results demonstrated the clinical utilities of the patient-specific features learned from the embedding algorithms. The latent patient features obtained via the embedding process enabled direct applications of other machine learning algorithms. Future work will focus on utilising the temporal information within EHR and extending EHR embedding algorithms to develop personalised patient journey predictions.

Arrears behavior prediction of power users based on BP neural network and multi-scale feature learning: a refined risk assessment framework

This study aims to develop an efficient model to predict the arrears behavior of electricity users by integrating multi-scale feature learning with a backpropagation (BP) neural network. The goal is to provide accurate early warning systems and enhanced risk management tools for power companies. The BP neural network algorithm adjusts weights to minimize prediction errors, while multi-scale feature learning captures the diversity and regularity of user behavior by extracting data from various time dimensions, such as daily, weekly, and monthly intervals. First, electricity usage and weather data from the UMass Smart Dataset are preprocessed, including steps such as data cleaning, standardization, and normalization. Next, features are extracted across three time scales—daily, weekly, and monthly. These features are then input into the BP neural network model using the multi-scale feature learning method. A hierarchical neural network structure is designed to address the characteristics of different scales in distinct layers. Key model parameters are optimized, and a sensitivity analysis is conducted. The experimental results demonstrate that the BP neural network model incorporating multi-scale features outperforms traditional BP neural network models and other control models in several evaluation metrics. Specifically, the Gini coefficient is 0.55, the Kolmogorov-Smirnov statistic is 0.60, the Matthews correlation coefficient is 0.45, and specificity is 0.82. These results indicate that the proposed method offers significant improvements in capturing user behavior patterns and enhancing prediction accuracy. The study concludes that the effective fusion of multi-scale features not only enhances the model’s prediction performance but also strengthens its generalization ability. This method provides an advanced risk management tool for power companies, helping to increase the operational efficiency of smart grids and encouraging further research toward greater intelligence in the field.

Intelligent target recognition method for infrared/laser composite fuze

Abstract In the context of complex environment, the intelligent target recognition method for infrared/laser composite imaging fuze encounters the following challenges: 1) The dearth of infrared/laser composite imaging fuze data hampers the advancement and implementation of deep learning technology in fuze target recognition; 2) The extant infrared/laser composite imaging fuze target recognition methods exhibit suboptimal recognition accuracy in complex terrain and ground conditions, which predisposes them to early or false explotions. To address the aforementioned issues, we have constructed infrared/laser composite imaging target and background dataset in diverse scenes, terrains and features, which provides effective data support for the intelligent recognition of composite imaging fuze. Additionally, we have designed a cross-modal fusion attention mechanism and an invariant feature learning method. The method facilitates the fusion of multimodal complementary information and the intelligent recognition of multi-scale and multi-morphology targets in complex scenes. Furthermore, they enhance the accuracy of the composite imaging fuze, enabling it to adapt to complex terrains. This enhances the combat effectiveness of composite imaging fuze in complex environments, while also offering technical support for the intelligentization of target recognition systems in future ballistic fuze.

Sparse Spike Feature Learning to Recognize Traceable Interictal Epileptiform Spikes.

Interictal epileptiform spikes (spikes) and epileptogenic focus are strongly correlated. However, partial spikes are insensitive to epileptogenic focus, which restricts epilepsy neurosurgery. Therefore, identifying spike subtypes that are strongly associated with epileptogenic focus (traceable spikes) could facilitate their use as reliable signal sources for accurately tracing epileptogenic focus. However, the sparse firing phenomenon in the transmission of intracranial neuronal discharges leads to differences within spikes that cannot be observed visually. Therefore, neuro-electro-physiologists are unable to identify traceable spikes that could accurately locate epileptogenic focus. Herein, we propose a novel sparse spike feature learning method to recognize traceable spikes and extract discrimination information related to epileptogenic focus. First, a multilevel eigensystem feature representation was determined based on a multilevel feature representation module to express the intrinsic properties of a spike. Second, the sparse feature learning module expressed the sparse spike multi-domain context feature representation to extract sparse spike feature representations. Among them, a sparse spike encoding strategy was implemented to effectively simulate the sparse firing phenomenon for the accurate encoding of the activity of intracranial neurosources. The sensitivity of the proposed method was 97.1%, demonstrating its effectiveness and significant efficiency relative to other state-of-the-art methods.

Recent kernel methods for interacting particle systems: first numerical results

Abstract Interacting particle systems (IPSs) are a very important class of dynamical systems, arising in different domains like biology, physics, sociology and engineering. In many applications, these systems can be very large, making their simulation and control, as well as related numerical tasks, very challenging. Kernel methods, a powerful tool in machine learning, offer promising approaches for analyzing and managing IPS. This paper provides a comprehensive study of applying kernel methods to IPS, including the development of numerical schemes and the exploration of mean-field limits. We present novel applications and numerical experiments demonstrating the effectiveness of kernel methods for surrogate modelling and state-dependent feature learning in IPS. Our findings highlight the potential of these methods for advancing the study and control of large-scale IPS.

Combining multi-level feature extraction algorithm with residual graph convolutional neural network for partial discharge detection

Partial discharge (PD) identification is critical for the insulation diagnosis of cable conductors; however, there is still scope for enhancement in the existing adaptive extraction capabilities and feature utilization for PD signals. In this context, this paper introduces a method that integrates a one-dimensional convolutional neural network (1D-CNN) with a residual graph convolutional neural network (ResGCN) to recognize PD signals. Noise analysis is performed using various combinations of mother wavelets, and Bayesian optimization is employed to mitigate background noise. Key features of PD signals are progressively abstracted through a 1D-CNN-based multilevel automatic feature learning method, while signal timing attributes are maintained to minimize manual intervention. The graph data is constructed using the signal feature matrix and the signal timing feature similarity matrix. This is followed by the development of a ResGCN utilizing a graph attention mechanism to integrate node feature information and the topology of the PD graph data. This approach aims to fully exploit the correlation between local regions of the feature space and the temporal numerical properties of the signals. Additionally, it jointly optimizes feature extraction and model classification to facilitate adaptive diagnosis. The method is validated with extensive real experimental data obtained from medium voltage overhead power lines. It demonstrates exceptional performance and practicality, achieving an accuracy rate of 97.3% and a recognition rate of 96.1% for PD samples, thus offering reliable theoretical support for effective PD detection.

Design of model fusion learning method based on deep bidirectional GRU neural network in fault diagnosis of industrial processes

This paper proposes an end-to-end model fusion feature learning method based on deep bidirectional gated recurrent unit (MCNN-DBiGRU) for fault diagnosis in industrial processes. First, a feature-aligned multi-scale feature extraction model (MCNN) is designed by analyzing the working principles of convolutional and pooling layers of convolutional neural networks. Secondly, a deep bidirectional mechanism is proposed to better extract the time series features in the process data. This mechanism makes the recurrent neural network not only present the forward processing input features from the past to the future, but also the reverse processing from the future to the past. By integrating these features, the diagnostic performance of the network model is improved. To verify that the proposed model has effective diagnostic accuracy for fault diagnosis, we conduct simulation experiments on the Tennessee-Eastman (TE) process and a chemical coking furnace, and compare with several conventional network models. In the end, not only the effectiveness of the model is proved, but it is also confirmed that the model is superior to other conventional neural networks in both diagnostic accuracy and feature robustness.

Unbiased Feature Learning with Causal Intervention for Visible-Infrared Person Re-Identification

Visible-infrared person re-identification (VI-ReID) aims to match individuals across different modalities. Existing methods can learn class-separable features but still struggle with modality gaps within class due to the modality-specific information, which is discriminative in one modality but not present in another (e.g., a black striped shirt). The presence of the interfering information creates a spurious correlation with the class label, which hinders alignment across modalities. To this end, we propose an Unbiased feature learning method based on Causal inTervention for VI-ReID from three aspects. Firstly, through the proposed structural causal graph, we demonstrate that modality-specific information acts as a confounder that restricts the intra-class feature alignment. Secondly, we propose a causal intervention method to remove the confounder using an effective approximation of backdoor adjustment, which involves adjusting the spurious correlation between features and labels. Thirdly, we incorporate the proposed approximation method into the basic VI-ReID model. Specifically, the confounder can be removed by adjusting the extracted features with a set of weighted pre-trained class prototypes from different modalities, where the weight is adapted based on the features. Extensive experiments on the SYSU-MM01 and RegDB datasets demonstrate that our method outperforms state-of-the-art methods. Code is available at https://github.com/NJUPT-MCC/UCT .

An Analytical Approach for IGBT Life Prediction Using Successive Variational Mode Decomposition and Bidirectional Long Short-Term Memory Networks

The precise estimation of the operational lifespan of insulated gate bipolar transistors (IGBT) holds paramount significance for ensuring the efficient and uncompromised safety of industrial equipment. However, numerous methodologies and models currently employed for this purpose often fall short of delivering highly accurate predictions. The analytical approach that combines the Pattern Optimization Algorithm (POA) with Successive Variational Mode Decomposition (SVMD) and Bidirectional Long Short-term Memory (BiLSTM) network is introduced. Firstly, SVMD is employed as an unsupervised feature learning method to partition the data into intrinsic modal functions (IMFs), which are used to eliminate noise and preserve the essential signal. Secondly, the BiLSTM network is integrated for supervised learning purposes, enabling the prediction of the decomposed sequence. Additionally, the hyperparameters of BiLSTM and the penalty coefficients of SVMD are optimized utilizing the POA technique. Subsequently, the various modal functions are predicted utilizing the trained prediction model, and the individual mode predictions are subsequently aggregated to yield the model’s definitive final life prediction. Through case studies involving IGBT aging datasets, the optimal prediction model was formulated and its lifespan prediction capability was validated. The superiority of the proposed method is demonstrated by comparing it with benchmark models and other state-of-the-art methods.

Comparison of feature learning methods for non-invasive interstitial glucose prediction using wearable sensors in healthy cohorts: a pilot study

BackgroundAlterations in glucose metabolism, especially the postprandial glucose response (PPGR), are crucial contributors to metabolic dysfunction, which underlies the pathogenesis of metabolic syndrome (MetSyn). Personalized low-glycemic diets (PLGD) have shown promise in reducing postprandial glucose spikes. However, current methods such as invasive continuous glucose monitoring (CGM) or multi-omics data integration to assess PPGR come with limitations, including cost and invasiveness, hindering widespread adoption in primary disease prevention. Our aim is to assess machine-learning algorithms for predicting individual PPGR using non-invasive wearable data, thereby circumventing the limitations associated with existing approaches. By identifying the most accurate model, we seek to provide a more accessible and efficient method for managing glucose metabolism. MethodsThis data-driven analysis utilized the experimental dataset from the SENSE (”Systemische Ernährungsmedizin”) study. Healthy participants used an Empatica E4 wristband and Abbott Freestyle Libre 3 CGM for ten days. Blood volume pulse (BVP), electrodermal activity (EDA), heart rate (HR), skin temperature (TEMP), and the corresponding CGM values were measured. Subsequently, four data-driven deep learning (DL) model-convolutional neural network (CNN), lightweight transformer (LTF), long short-term memory with attention (LSTM-A), and Bi-directional LSTM (BiLSTM)-were implemented and compared to determine the potential of DL in predicting interstitial glucose levels without involving food and activity logs. ResultsThe proposed BiLSTM achieved the best interstitial glucose prediction performance, with an average root mean squared error (RMSE) of 13.42 mg/dL, an average mean absolute percentage error (MAPE) of 0.12, and only 3.01% values falling within area D in Clarke error grid analysis (CEGA), incorporating the leave-one-out cross-validation (LOOCV) strategy for a five-minute prediction horizon. ConclusionThe findings of this study may demonstrate the feasibility of transferring knowledge gained from invasive glucose monitoring devices to non-invasive approaches. Furthermore, it emphasizes the promising prospects of combining deep learning with wearable technologies to predict glucose levels in healthy individuals.

Semantic Segmentation of Point Cloud Scene via Multi-Scale Feature Aggregation and Adaptive Fusion

Point cloud semantic segmentation is a key step in 3D scene understanding and analysis. In recent years, deep learning–based point cloud semantic segmentation methods have received extensive attention from researchers. Multi-scale neighborhood feature learning methods are suitable for inhomogeneous density point clouds, but different scale branching feature learning increases the computational complexity and makes it difficult to accurately fuse different scale features to express local information. In this study, a point cloud semantic segmentation network based on RandLA-Net with multi-scale local feature aggregation and adaptive fusion is proposed. The designed structure can reduce computational complexity and accurately express local features. The mean intersection-over-union is improved by 1.1% on the SemanticKITTI data set with an inference speed of nine frames per second, while the mean intersection-over-union is improved by 0.9% on the S3DIS data set, compared with RandLA-Net. We also conduct ablation studies to validate the effectiveness of the proposed key structure.

JFDI: Joint Feature Differentiation and Interaction for domain adaptive object detection

In unsupervised domain adaptive object detection, learning target-specific features is pivotal in enhancing detector performance. However, previous methods mostly concentrated on aligning domain-invariant features across domains and neglected integrating the specific features. To tackle this issue, we introduce a novel feature learning method called Joint Feature Differentiation and Interaction (JFDI), which significantly boosts the adaptability of the object detector. We construct a dual-path architecture based on we proposed feature differentiate modules: One path, guided by the source domain data, utilizes multiple discriminators to confuse and align domain-invariant features. The other path, specifically tailored to the target domain, learns its distinctive characteristics based on pseudo-labeled target data. Subsequently, we implement an interactive enhanced mechanism between these paths to ensure stable learning of features and mitigate interference from pseudo-label noise during the iterative optimization. Additionally, we devise a hierarchical pseudo-label fusion module that consolidates more comprehensive and reliable results. In addition, we analyze the generalization error bound of JFDI, which provides a theoretical basis for the effectiveness of JFDI. Extensive empirical evaluations across diverse benchmark scenarios demonstrate that our method is advanced and efficient.

Adaptive Attention-Enhanced Yolo for Wall Crack Detection

With the advancement of social life, the aging of building walls has become an unavoidable phenomenon. Due to the limited efficiency of manually detecting cracks, it is especially necessary to explore intelligent detection techniques. Currently, deep learning has garnered growing attention in crack detection, leading to the development of numerous feature learning methods. Although the technology in this area has been progressing, it still faces problems such as insufficient feature extraction and instability of prediction results. To address the shortcomings in the current research, this paper proposes a new Adaptive Attention-Enhanced Yolo. The method employs a Swin Transformer-based Cross-Stage Partial Bottleneck with a three-convolution structure, introduces an adaptive sensory field module in the neck network, and processes the features through a multi-head attention structure during the prediction process. The introduction of these modules greatly improves the performance of the model, thus effectively improving the precision of crack detection.

Learning sparse feature representation for blind quality assessment of night-time images

Capturing Night-Time Images (NTIs) with high-quality is quite challenging for consumer photography and several practical applications. Thus, addressing the quality assessment of night-time images is urgently needed. Since there is no available reference image for such images, Night-Time image Quality Assessment (NTQA) should be done blindly. Although Blind natural Image Quality Assessment (BIQA) has attracted a great deal of attention for a long time, very little work has been done in the field of NTQA. Due to the capturing conditions, NTIs suffer from various complex authentic distortions that make it a challenging field of research. Therefore, previous BIQA methods, do not provide sufficient correlation with subjective scores in the case of NTIs and special methods of NTQA should be developed. In this paper we conduct an unsupervised feature learning method for blind quality assessment of night-time images. The features are the sparse representation over the data-adaptive dictionaries learned on the image exposure and gradient magnitude maps. Having these features, an ensemble regression model trained using least squares gradient boosting scheme predicts high correlated objective scores on the standard datasets.

Additive dynamic Bayesian networks for enhanced feature learning in soft sensor modeling

Due to the advantages of indicating variable structure and efficient reasoning, Bayesian Networks (BN) have been widely used in data-driven soft sensor applications. However, restricted to linear and conditional Gaussian property, BN-based soft sensors rarely achieve high prediction accuracy. In this paper, an ensemble learning framework – additive dynamic Bayesian networks (ADBN) is proposed for enhanced feature learning, in which Dynamic Bayesian networks are used to learn the conditional independencies among variables and construct feature sets for the following base learners. Additional DBNs are constructed upon the residual information from the past model, to carry out feature learning to fit the residuals. The procedure is repeated and a termination rule from the perspective of feature learning is proposed to end this process, and thus the model complexity can be well restricted. The proposed method is validated on two actual industrial cases. It reveals that the ADBN feature learning method has obtained great improvements. Compared to the single DBN feature engineering method, the root mean square error (RMSE) performance has been improved by 20% and 13% on the two cases, respectively.

Restoration towards decomposition: A simple approach for domain generalization

Domain generalization (DG) aims to train a model capable of generalizing to unseen target domains by utilizing data from multiple disjoint source domains. Previous domain generalization methods primarily utilize data augmentation and domain-invariant feature learning. However, since the target domain is unknown, data augmentation methods struggle to generate images or features that closely resemble the target domain. In addition, existing domain-invariant feature learning methods are incapable of achieving complete decoupling, resulting in misalignment of feature distribution between the source and target domains. Hence, in this work, we propose a new perspective to address domain generalization by focusing on learning the ability to transform the target domain visual representations into those of the source domain. If the style of the unknown target domain can be transformed into the style of the known source domain, the adverse effects of domain shift can be effectively mitigated. Following this line of thought, we have developed a meta-learning task named “feature restoration” aimed at training a model to acquire this capability. Specifically, feature restoration involves replacing the domain-specific components within the visual representations of the target domain with those of the source domain. Experimental results on five DG benchmarks reveal that our method can achieve state-of-the-art performance. Ablation studies and visualization results further demonstrate the rationality and effectiveness of our design.

RCFI-Net: A reliable correspondences evaluation and feature interaction network for fast and accurate point cloud registration

3D point cloud registration is a crucial task in computer vision, robotics, and safeguarding cultural artifacts in the digital realm. However, maintaining a balance between efficiency and accuracy during the 3D point cloud registration process continues to be a challenge. To solve this issue, a Reliable Correspondences Evaluation and Feature Interaction Network called RCFI-Net has been proposed for fast and precise registration. Our model use a two-stage approach to improve registration efficiency. In the first stage, a Transformer with position encoding network is employed to reinforce point features. Overlapping masks are learned based on the attention mechanism to identify overlapping areas, followed by the sampling of interest points that have high scores in these areas to accelerate registration. In the second stage, non-distinguishing points are eliminated, and a triangulated descriptor is introduced to further differentiate inliers and outliers and find reliable correspondences. Our proposed method has been evaluated on both public datasets and real Terracotta Warriors data, with results showing that it outperforms traditional and feature-learning methods in terms of accuracy, efficiency, and robustness.

An End-to-End Geometric Characterization-aware Semantic Instance Segmentation Network for ALS Point Clouds

Abstract. Semantic instance segmentation from scenes, serving as a crucial role for 3D modelling and scene understanding. Conducting semantic segmentation before grouping instances is adopted by the existing state-of-the-art methods. However, without additional refinement, semantic errors will fully propagate into the grouping stage, resulting in low overlap with the ground truth instance. Furthermore, the proposed methods focused on indoor level scenes, which are limited when directly applied to large-scale outdoor Airborne Laser Scanning (ALS) point clouds. Numerous instances, significant object density and scale variations make ALS point clouds distinct from indoor data. In order to address the problems, we proposed a geometric characterization-aware semantic instance segmentation network, which utilized both semantic and objectness score to select potential points for grouping. And in point cloud feature learning stage, hand-craft geometry features are taken as input for geometric characterization awareness. Moreover, to address errors propagated from previous modules after grouping, we have additionally designed a per-instance refinement module. To assess semantic instance segmentation, we conducted experiments on an open-source dataset. Additionally, we performed semantic segmentation experiments to evaluate the performance of our proposed point cloud feature learning method.

New Multi-View Feature Learning Method for Accurate Antifungal Peptide Detection

New Multi-View Feature Learning Method for Accurate Antifungal Peptide Detection