Search

Identifying the abnormal samples from a data set and determining their type are two key tasks of anomaly detection. However, the existing anomaly detection algorithms are generally faced with the defects of weak generalization ability and insufficient interpretation, the core reason of which is that they cannot mine specific features for different abnormal types. In this paper, a new anomaly detection algorithm aiming at feature selection for different abnormal types is developed. Inspired by self-supervised learning, we take the stationarity of variance changes of abnormal score similarity as a pretext task and combine it with wrapped search method. Then, the features and the corresponding parameters for different abnormal types can be screened to apply to the downstream task of anomaly integration detection. To verify the efficiency of the new algorithm, we conduct two sets of experiments to compare the new algorithm with 11 classical anomaly detection and 3 clustering anomaly detection algorithms on the data sets WDBC, WPBC and Wilt from DAMI database with the evaluation measures P@n , Adj- P@n , AP, Adj-AP and AUC. The experiment results show that, both in the identification and classification on abnormal samples, all performance measures of the new algorithm are explicitly better than that of the contrastive algorithms. Also, we apply the new algorithm to the Chinese auto insurance market, and find that the results can help managers to identify the main patterns of fraudulent claims and to summarize the feature combinations of fraud behaviors. In general, the new algorithm developed in this paper has the following advantages compared with traditional algorithms: 1) It can directly capture abnormal features and realize effective recognition of abnormal types, which effectively bridge the gap between abnormal judgement and feature screening. 2) Automatic screening of abnormal features can be completed under the condition of self-updating learning optimal strategy. 3) Only a few features are extracted from all features to reveal the abnormal characteristics, which significantly improves the interpretability and generalization ability of the algorithm and its results. In a word, through the novel self-supervised design method, feature screening is skillfully integrated into the anomaly detection task, which may provide a new way for anomaly detection research. [ABSTRACT FROM AUTHOR]

The performance of object detection networks is often limited by the depth of the feature extraction network. Increasing network parameters may yield limited improvements in the detection system's performance. Additional careful designs of network details are necessary, but they can significantly increase training difficulty. This paper introduces a novel object detection method that utilizes generative adversarial training. Our approach focuses on minimizing the EM distance (Wasserstein distance) of the feature distribution as the primary training objective. We enhance the image features so that training of GAN (Generative Adversarial Networks) yields a feature distribution that exceeds that of the original dataset, obtaining a better feature extraction network. A new loss function is also added to the adversarial training process to ensure stable improvement of the detector. A comparative experiment conducted with the original CenterNet network on MS COCO (Microsoft Common Objects in COntext) 2017 reveals that the generative adversarial training method significantly improves the average precision for most of the examined backbone networks. Among the four backbone networks employed in the experiments, the mean improvement in network AP (Average Precision) values ranged from 0.3 to 0.9, demonstrating their success with minimal training efforts. Moreover, none of the four backbone networks experienced an increase in network parameters during inference. Experimental results indicate that the proposed architecture effectively enhances the network's feature extraction capability without compromising speed during inference. [ABSTRACT FROM AUTHOR]

Feature extraction is one of the fundamental problems in pattern recognition research. For image recognition, extracting effective image features is the key to accomplish the recognition task. In this paper, a partial differential equations-based multi-rank two-dimensional canonical correlation analysis (PDEs-MR2DC2A) is proposed for multi-view feature extraction and pattern classification. Unlike most of the previous researches on multi-view algorithms that work directly on the original 2D representation, in our approach, we first utilize the evolution process of PDEs to extract the feature matrix of per-view. In addition, we employ multi-rank left and right projecting matrices to maximize the correlation. The computational complexity of PDEs-MR2DC2A is also analyzed. To evaluate the effectiveness of the proposed algorithm, we conducted a series of performance comparisons with some existing methods on several popular datasets. The experimental results showed that our proposed algorithm performed very well on these datasets and outperformed the existing related methods on some metrics. • A novel multi-rank multi-view algorithm based on PDEs is proposed. • This approach can extract rotation and translation invariant features. • Numerous experimental results demonstrate the effectiveness of our proposed method. [ABSTRACT FROM AUTHOR]

• We focused on deep learning (DL), PAD and multimodal finger vein recognition. • DL, PAD and multimodal approaches are comprehensively studied. • A comparative study is presented for reviewed paper based on performance. • Existing challenges, problems in DL, PAD and multimodal approaches are discussed. • Prospect for future work in this area for stable finger vein recognition system. Finger vein recognition biometric trait is a significant biometric modality that is considered more secure, reliable, and emerging. This article presents a review to focus on the recent research landscape in biometric finger vein recognition systems. This article focuses on manuscripts related to keywords 'Finger Vein Authentication System', 'Anti-spoofing or Presentation Attack Detection', 'Multimodal Biometric Finger Vein Authentication' and their variations in four main digital research libraries such as IEEE Xplore, Springer, ACM, and Science Direct. The final set of articles is divided into three main categories: Deep Learning (DL) based finger vein recognition, Presentation Attack Detection (PAD), and Multimodal-based finger vein authentication system. Deep learning-based finger vein recognition techniques are further sub-divided into pre-processing (Quality assessment and enhancement) based, feature extraction based, and feature extraction and recognition based schemes. Presentation attack detection methods are sub-divided into software-based and hardware-based approaches. Multimodal-based finger vein biometric system is sub-categorized into feature level fusion, matching level fusion, and hybrid fusion methods. The authors have studied the problem of the recent algorithm and their solution related to finger vein biometric system from the recent literature. Performance analysis and selected the best promising research work from the mentioned studies are also presented. Finally, open challenges, opportunities, and suggested solutions related to deep learning, PAD, and Multimodal based finger vein recognition systems have been discussed in the discussion section. This work would be helpful to the developers, company users, researchers, and readers to get insight into the importance of such technology and the recent problem faced by finger vein authentication technology with respect to deep learning and multimodal systems. [ABSTRACT FROM AUTHOR]

The mortality rate due to cardiac abnormalities is enormous, making the development of wearables to monitor functioning of the heart of paramount importance. In this paper, wepresent a resource efficient and low power architecture using Integer Haar Wavelet Transform for the complete delineation of ECG signal. The novelty of the algorithm lies in the use of single scale wavelet coefficients to delineate P-QRS-T features making it computationally simple. The proposed architecture is implemented using Xilinx FPGA ZedBoard Zynq™ − 7000 platform, and utilises only 4.38% of the available resources. It is synthesised using 180 nm CMOS technology consuming 0.88 μ W power, making it area as well as power-efficient for the wearable IoT healthcare devices. • Extracting features of ECG (P-QRS-T) using single scale wavelet coefficients. • Integer Haar wavelet is utilised to avoid floating point arithmetic reducing computational complexity. • Architecture developed operates in low power and utilise minimum hardware resources. • Architecture can be used in low power wearable IoT healthcare applications. [ABSTRACT FROM AUTHOR]

Feature extraction networks play a crucial role in classification algorithms. However, most convolutional neural networks use the form of residual block stacking to extract downstream features of images, and this simple form is not sufficient to obtain better classification performance. Therefore, this paper proposes a YOLOX detection framework based on an improved ConvNeXt which incorporates many advantages of Swin-Transformer with efficient classification performance. To address the problems of unstable IOU convergence and inaccurate regression coordinates of the algorithm, this paper adopts Manhattan distance as the new penalty term, and the new loss function combines the advantages of Euclidean distance and Manhattan distance, which has no gradient explosion in the early training period and can drop smoothly to a stable point in the late training period. The proposed algorithm is robust to negative effects such as dust, light, shadows, and occlusions, and is well-suited for masked-face detection in industrial and public scenarios. The proposed model is tested on the VOC dataset and the AIZOO open-source masked-face dataset, obtaining 5.58% AP improvement on the VOC dataset and reaching 96.665% AP on the masked-face dataset, which is better than other algorithms. • Designing a deep neural network to use for masked-face detection. • Anchor-free model is used to avoid time-consuming post-processing. • The improved ConvNeXt network improves classification performance. • Adopting Manhattan distance-based localization loss makes training stable. • The Mosaic data augmentation strategy alleviates the overfitting problem. [ABSTRACT FROM AUTHOR]

Segmentation of bowel sounds (BS) events is a significant task of automatic BS monitoring. Recently, deep learning (DL) has been utilized to realize the segmentation of BS events. However, most researchers treat BS segmentation as a traditional classification problem, which causes that more precise locations for the occurrence of BS events are unable to obtained. Besides, the performance of segmentation of BS events is easily affected by thresholds of model output in practical applications. To tackle these issues, in this paper, a lightweight DL-based BS segmentation algorithm is proposed. The one-dimensional convolution layers and bidirectional gate recurrent unit (GRU) layers are adopted to enhance the ability of feature extraction. A loss function named shape loss is proposed to reduce the sensitivity of model to thresholds. Moreover, a portable BS monitoring device is developed to realize data acquisition, display and algorithm deployment. Based on this device, experiments on human and rats are conducted to verify the effectiveness of our proposed approach. Experimental results shows that the proposed method outperforms other comparison methods with the higher f1-scores and lower sensitivity to thresholds. [ABSTRACT FROM AUTHOR]

The continuous malicious software (malware) attacks on smartphones pose a serious threat to the security of users, especially the dominant platform Android. Data-driven methods based on machine learning algorithms are a promising way to defend against that. In this paper, we explore the limitations of this kind of methods in improving the performance of malware detection, e.g., considering each feature in isolation and relying on balanced class data, and propose a multi-model ensemble framework MEFDroid by combining individual predictors, where hybrid deep learning based feature extraction methods are adopted to learn meaningful features from raw data. Besides, a novel fusion scheme is exploited to fuse multi-level representations and mine their correlations to maximize the utilization of original information. To evaluate the effectiveness of the MEFDroid framework, we conduct a step-by-step model justification experiment to investigate how the proposed algorithms (i.e., ESAES, EDAES and EDAFS) enhance the malware detection performance gradually. We also compare the proposed algorithms with classical machine learning methods, conventional sampling solutions for the imbalance problem. Besides, we investigate the reliability of our proposed methods using another public dataset. Our extensive experimental results demonstrate that the target model EDAFS in MEFDroid outperforms others in terms of most metrics, which means that it is an effective alternative solution to detect Android malware. [ABSTRACT FROM AUTHOR]

To segment railway track regions in real-time for intrusion detection and improving security, this paper proposes an efficient railway track region segmentation network (ERTNet) based on the encoder–decoder architecture. Firstly, to ensure the lightweight of the encoder, depthwise convolution and the channel shuffle are utilized to construct sandglass-type feature extraction unit. Secondly, a feature-matching-based cross-fusion decoder is utilized to fuse deep and shallow feature maps. Thirdly, the knowledge distillation is employed with large-scale Deeplab v3+ as the teacher model to improve performance. Additionally, a loss function is proposed to penalize pixel points with large offsets. Finally, the ERTNet is validated on the self-built dataset, achieving an MIoU (Mean Intersection over Union) of 92.4% , which is 5.22% improvement over the benchmark model. ERTNet achieves a balance between segmentation accuracy and computational efficiency, requiring only 0.5 M parameters and 0.92 G FLOPs (Floating Point Operations). [Display omitted] • Visual perception is utilized to identify foreign object intrusion into railroad. • Deep learning based ERTNet model is proposed for track region pixel extraction. • ERTNet is built with lightweight neural network and cross-fusion decoder. • Optimizing training modes with knowledge distillation and new loss function. • ERTNet fits real-time needs and helps to improve the safety of rail construction. [ABSTRACT FROM AUTHOR]

The occurrence of production accidents can be effectively reduced by monitoring workers' operating behaviors in real time. However, most of the monitoring tasks are currently performed by the monitoring personnel, which takes up a lot of manpower and material resources. To solve this problem, a YOLOv5-SFE algorithm is proposed in this paper for real-time detection and recognition of workers' operating behaviors. The YOLOv5-SFE algorithm makes the following contributions: (1) During data preprocessing, a hash sampling algorithm is used to extract frames with low similarity. (2) A feature enhancement module is designed and integrated into YOLOv5 to distinguish between valid and invalid information. (3) A convolution-based spatio-temporal feature fusion module is designed and is inserted after the extraction of spatial features to extract the temporal features between multiple frames. The videos of workers' operating behaviors are from factories' industrial scene. The improved algorithm in this paper was trained and tested on the dataset. Compared with the original algorithm, the accuracy of the algorithm improves from 89.3% to 94.7%, the recall improves from 81.5% to 90.8%, and the mean average precision(mAP) improves from 88.2% to 92%. The results show that the improved algorithm is able to accurately detect and recognize workers' operating behaviors in real time, thereby improving the safety of the production process. [ABSTRACT FROM AUTHOR]

Non-contact identification of cattle is the basis for cattle management, insurance claims, and traceability of milk sources and meat products. In order to lower the threshold of deep learning technology in practical application and improve the recognition accuracy, this paper uses the localization and classification paradigm, fills the image and proposes the LKA (Locate Key Area) module, which further improves the identification accuracy of cattle without using any labeling information. To avoid the impact of different collection devices, we use three brands of mobile phones and two cameras to collect data. Due to the fact that CNN features do not have scale invariance, the image is filled as a square, and the filling value is discussed. Finally, (0,0,0) is selected for filling. In order to reduce the labeling of new data, the face, torso, and body of the cattle are first detected, and then automatically trimmed to build a classification dataset. In order to further improve the accuracy of classification, LKA module is proposed. This module locates key areas based on the results of class activation map and uses shallow features for monitoring. Thereby completing fine-grained feature extraction without supervision information. Finally, the features of different parts are adjusted for fusion recognition. Experiments were conducted on the head, torso, and whole of cattle. Based on the ResNet50 network architecture, after filling the image into a square, the recognition accuracy has been improved by 0.30%, 0.21%, and 0.43%, respectively. After further adding LKA modules, the accuracy rates were improved by 0.59%, 0.59%, and 0.45%, respectively. In the fusion experiment, ResNet50_LKA and MobileNetV3_LKA achieves 100% accuracy. Experiments have realized dataset construction and fine-grained feature extraction without labeled information, and achieved high recognition accuracy, proving the effectiveness of our method. [ABSTRACT FROM AUTHOR]

Robot object detection is important for the realisation of robot intelligence. Currently, deep learning-based object detection algorithms are used for robotic object detection. However, it faces some challenges in practical applications, such as the fact that robots frequently use resource-constrained devices, resulting in detection algorithms with long computation times and undesired detection rates. In order to address these concerns, this paper proposes a lightweight object detection algorithm for robots with an improved YOLOv5. To reduce the amount of processing required for feature extraction and increase the speed of detection, the C3Ghost and GhostConv modules have been introduced into the YOLOv5 backbone. The DWConv module was used in conjunction with the C3Ghost module in the YOLOv5 neck network to further reduce the number of model parameters and maintain accuracy. The CA (Coordinated Attention) module is also introduced to improve the extraction of features from detected objects and suppress irrelevant features, thus improving the algorithm's detection accuracy. To verify the performance of the method, we tested it with a self-built dataset (4561 robot images in total) and the PascalVOC dataset respectively. The results show that compared with the YOLOv5s on the self-built dataset, the algorithm has a 54% decrease in FLOPs and a 52.53% decrease in the number of model parameters without a decrease in mAP (0.5). The effectiveness and superiority of the algorithm is demonstrated through case studies and comparisons. • Lightweight C3Ghost and GhostConv modules are introduced in the YOLOv5 backbone network to achieve model compression and maintain detection accuracy and speed. • C3Ghost and DWConv modules are introduced in YOLOv5 neck network to further reduce model parameters and improve the speed of feature fusion. • CA (Coordinated attention) module is also introduced to enhance the extraction of relevant features and suppress irrelevant features to improve the detection accuracy of the algorithm. • To demonstrate the algorithm's ability to solve real-world problems, we produced a dataset for the object detection task in the RoboMasterAI challenge hosted by DJI and experimentally verified that our proposed object detection algorithm is effective. [ABSTRACT FROM AUTHOR]

Accurately detect vehicles or pedestrians from 3D point clouds (3D object detection) is a fast developing research topic in autonomous driving and other domains. The fundamental component for feature extraction in 3D object detection is Set Abstraction (SA), which can downsample points while aggregating points to extract features. However, the current SA ignores the geometric and semantic properties of point clouds and may miss to detect remote small objects. In this paper, FocusSA is proposed, which consists two modules for enhancing useful feature extraction in the SA layer to improve 3D object detection accuracy. At first, Focused FPS (FocFPS) is proposed to evaluate the foreground and boundary scores of the points and reweighs the Furthest Point Sampling (FPS) using the evaluated scores to retain more contextual points in downsampling. Then a Geometry-aware Feature Extraction (GeoFE) module is proposed to add geometric information to enrich the awareness of geometric structure in feature aggregation. To evaluate the performances of the proposed methods, we conduct extensive experiments on three difficulty levels of Car class in KITTI dataset. The experimental results show that on "moderate" instances, our results outperform the state-of-the-art method by 1.08%. Moreover, FocusSA is easy to be plugged in popular architectures. [ABSTRACT FROM AUTHOR]

This aims to propose a collaborative filtering algorithm based on embedding representation and word embedding techniques, namely UI2vec. According to the joint feature extraction network designed in this paper, UI2vec embeds users and items on the potential space at the same time, and uses the item similarity between them to predict the user's content of interest. Then a generative model VUI2vec with more stable performance is proposed based on UI2vec, which maps users and items as independent Gaussian distributions and obtains the approximate posterior distribution of both by variational inference. The recommendation performance of UI2vec and VUI2vec is evaluated on TaFeng, Movielens, and Netflix datasets. The impact of important superparameters within the model on performance is investigated. The experimental results show that compared with the baseline model, the proposed methods performs consistently well. • UI2vec and VUI2vec as collaborative filtering-based recommendation systems. • It is based on embedding representation of users or items. • Performance is competitive with the state of the art on real-world datasets. • Results encourage integrating collaborative filtering with deep learning. [ABSTRACT FROM AUTHOR]

The Open Circuit (OC) faults occurring in switches of Multilevel Converters (MLC) may lead to undesirable operation of the converter. Therefore, fault detection and its localization in minimum time are necessary. This paper focuses on the fast fault detection algorithm based on the two samples technique and the fault localization algorithm using the Entropy of Wavelet Packets (EWP) as a feature. The EWP feature is used to classify and localize the OC faults in Insulated Gate Bipolar Transistors (IGBTs) of three-phase, three-level inverter using Support Vector Machine (SVM) based fault classification algorithm. The proposed technique can detect the fault in single IGBT and multiple IGBTs in a lesser time range of microseconds to 0.33 ms. It gives better performance and accuracy (99.70%) than previously proposed SVM algorithms, as the EWP-based feature extraction process used in this paper is simple and accurate with a less computational burden. [Display omitted] • An observer-based algorithm is proposed using two samples-based technique. • The Entropy of Wavelet Packets (EWP-SVM) technique is used for fault diagnosis. • The proposed algorithm can detect the faults in a single IGBT and multiple IGBTs. • The technique is faster and accurate than techniques available in the literature. • The comparison of detection time of different techniques is also discussed. [ABSTRACT FROM AUTHOR]

Detection transformer (DETR) combines convolutional neural network (CNN) with transformer, providing a more advanced idea. In this paper, an object detection model based on DETR is proposed for fire and smoke detection. Compared with other methods that based on deep learning, the proposed one simplifies the pipeline of detection and builds an end-to-end detector. At the same time, the original DETR model usually requires long training time and large amount of computation, resulting in relatively poor performance in detection speed and accuracy, and it shows to be not friendly for small or early fire detection. Therefore, when designing the proposed network model, a normalization-based attention module is added in the feature extraction stage to highlight the effective features, being beneficial to the process of encoding. A multiscale deformable attention is also used in the encoder–decoder structure, accelerating therefore, the process of convergence during the training of the model, which includes the enhancement of the detection of small objects. Also, considering the cost of computation, the number of layers in the encoder–decoder structure is redefined to reduce the complexity of the model, and that also reduces the requirements of the application equipment. Detailed experiments are conducted on three self-built datasets and two public video sets. The results show that, the proposed method has an excellent performance on all of the datasets considered here. [ABSTRACT FROM AUTHOR]

1 Creating a dataset for estimating depth from target foreground and background scene variations. 2 Creation of relationships between depth of person, foreground and background scene field. 3 Dealing with perspective errors in depth predictions while capturing still photographs of person along camera axial line. This paper proposes a new technique to achieve a person's depth from a single image by considering the target foreground and background scene variations in extreme weather conditions within 40 meters range. For this purpose series of images are captured on each person at successive intervals. The height, distance, foreground, and background features are extracted using an object detection deep learning framework. The obtained features are then subsequently trained by the Gradient Booster Regressor to predict the depth information. Furthermore, the algorithm is tested on various images and is validated with ground truth depth data. The findings presented in this paper attest to the reliability of the methodology used for depth estimation. [Display omitted] [ABSTRACT FROM AUTHOR]

• The novelty of the work presented in this paper is that a information fusion algorithm of the maximum multi-modal weight decision is proposed. The emotion recognition scheme is developed by using the multi-modal information fusion algorithm. • Another major contribution of this work is the innovative conversion of traditional EEG data into corresponding EEG topographic maps for the emotion recognition. • Besides, this paper builds a multi-scale feature extraction network model for the facial expression recognition. In view of the low accuracy of the traditional emotion recognition methods based on facial expressions, an emotion recognition method based on maximum weight multi-modal information fusion of electroencephalographs (EEGs) and facial expression information is proposed in this paper. First, the induced emotional EEG data is converted into the corresponding EEG topographic map data and sent to the convolutional network for training and outputting decision information. Second, the illumination compensation method is utilized to filter the noise of the face image data. Then, the face image data is trained in the multi-scale feature extraction network, and the decision information is output. Finally, aiming at the decision-level information fusion, a weighted fusion method is proposed in this paper for emotion recognition. Experimental tests show that the recognition accuracy of the multi-scale feature extraction network on the CK+ data set and Fer2013 data reached 94.4% and 72%, respectively. Simultaneously, the multi-modal information fusion method achieves 92.6% accuracy in emotion recognition. [Display omitted] The novelty of the work presented in this paper is that a fusion method of maximum weight decision information is proposed. Emotion recognition is realized by the weighted fusion of EEG information and facial expression information. Another major contribution of this work is the innovative conversion of traditional EEG data into corresponding state EEG topographic maps for processing and analysis. Besides, this paper builds a multi-scale feature extraction network for emotion recognition. The graphical abstract of this paper is as below. [ABSTRACT FROM AUTHOR]

• 3 layer neural networks outperform other ML algorithms in automated assessments. • In subjective assignments, deep neural networks can improve score prediction. • Word embeddings and linguistic feature results in more precise text mining models. • Extraction and selection of appropriate features can improve results. Machine learning (ML) has recently gained popularity in a variety of domains for automating tasks. This has also been used in the evaluation of student responses/ assignments. This paper presents a qualitatively enhanced methodology for automated score prediction of subjective assignments. To consider all the major aspects of a human grader, 21 linguistic features related to syntactical, grammatical, sentimental, and readability are analyzed from the assignments. This study makes use of the ASAP competition dataset from Kaggle. The evaluation metric used is Quadratic Weighted Kappa (QWK), which measures the agreement between the human graded score and the predicted score. The effect of appropriate feature selection has been observed using Mutual Information Regression. Four ML algorithms are investigated on identified linguistic features. 3 Layer Neural Network with feature selection performed well among all chosen ML algorithms with average QWK of 0.678. To include the benefits of deep learning, a new hybrid model (LF-BiLSTM-att-FS) is proposed that combines a higher level deep neural network (DNN) with the selected features. Pre-trained Glove embedding is used to include contextual information of the text. The proposed model results demonstrated an improvement in overall accuracy, with an average QWK value of 0.768. [ABSTRACT FROM AUTHOR]

Research about recommender systems emerges over the last decade and comprises valuable services to increase different companies' revenue. While most existing recommender systems rely either on a content-based approach or a collaborative approach, there are hybrid approaches that can improve recommendation accuracy using a combination of both approaches. Even though many algorithms are proposed using such methods, it is still necessary for further improvement. This paper proposes a recommender system method using a graph-based model associated with the similarity of users' ratings in combination with users' demographic and location information. By utilizing the advantages of Autoencoder feature extraction, we extract new features based on all combined attributes. Using the new set of features for clustering users, our proposed approach (GHRS) outperformed many existing recommendation algorithms on recommendation accuracy. Also, the method achieved significant results in the cold-start problem. All experiments have been performed on the MovieLens dataset due to the existence of users' side information. • We propose a recommendation system using graph-based features and Autoencoders. • We investigate and analyze each step of the method to find the best setup. • We examine the method with the most popular evaluation metrics and datasets. • Results show an improvement in comparison with basic and state of the art methods. [ABSTRACT FROM AUTHOR]

[Display omitted] • SentiWordNet lexicon is used for generating sentiment features of email data. • Probabilistic Latent Semantic Analysis algorithm is used for clustering. • The sentiment sequence of thread and thread size are identified. • Evaluated the proposed model using accuracy, precision, recall and F-measure. Email messaging is the most common way of providing effective communication between internauts. Consequently, the total sent and received emails count will be increased. But, the internaut can't remember all such emails. Even though email thread identification approaches give satisfactory benefits to the internauts, but they may fail to alert them for a cause to identify the sentiments behind an email thread. To address, this issue Probabilistic Latent Semantic Analysis clustering algorithm has been used in this paper to identify the email sentiment thread sequence. The sentiment and the thread sequence within the emails have been discovered as clustering sentiment polarity and temporal categories with the help of PLSA clusters. At the initial stage, we used three feature extraction methods, latent semantic analysis (LDA), bag of words (BoW), TF-IDF and SentiWordNet (SWN) lexicon for generating sentiment features of email. Next, Probabilistic Latent Semantic Analysis algorithm is used to form email clusters based on sentiment features. Thus, it helps to identify thread sentiment and sequence of sentiment threads. Email threads give a mechanism by which any user will be able to find out the sequence in the thread on the basis of sentiment analysis of email related to a specific set of communication during a specific time period. Various parameters evaluation measures have been considered in this work to evaluate the proposed model such as accuracy, precision, recall and F-measure, and the proposed algorithm is compared with other standard algorithms. Furthermore, a statistical test has also been performed. [ABSTRACT FROM AUTHOR]

Feature selection in real-world data mining problems is essential to make the learning task efficient and more accurate. Identifying the best feature selection algorithm, among the many available, is a complex activity that still relies heavily on human experts or some random trial-and-error procedure. Thus, the automated machine learning community has taken some steps towards the automation of this process. In this paper, we address the metalearning challenge of recommending feature selection algorithms by proposing a novel meta-feature engineering model. Our model considers a broad collection of meta-features that enable the study of the relationship between the dataset properties and the feature selection algorithm performance in terms of several criteria. We arrange the input meta-features into eight categories: (i) simple, (ii) statistical, (iii) information-theoretical, (iv) complexity, (v) landmarking, (vi) based on symbolic models, (vii) based on images, and (viii) based on complex networks (graphs). The target meta-features emerge from a multi-criteria performance measure, based on five individual performance indexes, that assesses feature selection methods grounded in information, distance, dependence, consistency, and precision measures. We evaluate our proposal using a recently developed framework that extracts the input meta-features from 213 benchmark datasets, and ranks the assessed feature selection algorithms, to fill in the target meta-features in meta-bases. This evaluation uses five state-of-the-art classification methods to induce recommendation models from meta-bases: C4.5, Random Forest, XGBoost, ANN, and SVM. The results showed that it is possible to reach an average accuracy of up to 90% applying our meta-feature engineering model. This work is the first to use an extensive empirical evaluation to provide a careful discussion of the strengths and limitations of more than 160 meta-features. These meta-features, while designed to aid the task of feature selection algorithm recommendation, can readily be employed in other metalearning scenarios. Therefore, we believe our findings are a valuable contribution to the fields of automated machine learning and data mining, as well as to the feature extraction and pattern recognition communities. • A novel meta-feature engineering model recommends feature selection algorithms. • The proposal obtains promising results from 213 datasets with hit rates of up to 90%. • Some simple, landmarking, image, and graph-based input meta-features highlighted. • A multi-criteria performance measure rigorously assesses candidate algorithms. • Chains of binary or multiclass classifiers can efficiently rank candidate algorithms. [ABSTRACT FROM AUTHOR]

Accurate prediction of the remaining useful life of lithium-ion (Li-ion) batteries is particularly important for their prognosis and health management. Therefore, a new feature extraction technique for extracting health indicators (HIs) characterizing the battery aging and a new improved extreme learning machine (ELM) algorithm for model training and prediction are proposed in this paper. Firstly, based on the measurable parameters, singular value decomposition (SVD) is used to extract the respective singular value as HIs, and then the Pearson correlation coefficient between each HI and capacity are calculated. Next, several HIs with high correlation coefficients are selected as the input of the model. Finally, the relationship model between HIs and capacity is constructed by using the improved ELM (OS-PELM) algorithm, and the final prediction results are obtained. Li-ion battery data from three different research institutions are adopted to verify the feasibility and reliability of the proposed method. Experiment results show that feature extraction technique and improved algorithm can not only extract features highly related to capacity, but also ensure the accuracy of prediction. The comparison with other algorithms further shows that the novel method is more accurate and competitive. • A technique called SVD is used to extract features. • OS-PELM is added online sequential learning based on PELM. • Battery data sets from three different institutions are considered. • The proposed method demonstrates high prediction accuracy, real-time and robustness. [ABSTRACT FROM AUTHOR]

• We design a new strong baseline method based on an adaptive representation algorithm, called obtaining sufficient features for logo (OSF-Logo), which aims to solve the logo deformation and multi-scale problem. • Our method adds a modulated deformable convolution module with adaptive offsets and amplitudes to more convolution layers in the stage of feature extraction. • A set of efficient feature sampling operations is introduced into FPN. Through feature reorganization of prediction kernel, the features of each layer can be fully integrated, which can adaptively obtain more comprehensive feature content, and effectively solve the deformation and multi-scale problem of logo images. • We conduct extensive experiment and evaluations to verify the effectiveness of our method on four different scale benchmark datasets. Our method has achieved good results. The proliferation of logo has driven research into multiple applications, like logo duration monitoring in advertising videos and logo infringement detection. Recently, logo research has attracted a rapt attention and keen interest from researchers. In these studies, logo detection is challenging due to the characteristics of the logo (like multi-scale, large-scale categories, viewpoint and part deformation etc.) and the complexity of its background. In this paper, we design a new strong baseline method based on an adaptive representation algorithm, called obtaining sufficient features for logo (OSF-Logo). The method aims to address the challenges of the multi-scale objects, large-scale categories, viewpoint logos and part deformation via introducing two modules. Specifically, we introduce a regulated deformable convolution module with offsets and amplitudes to more convolution layers in the stage of feature extraction. In addition, we add an up-sampling operator to FPN for aggregating information into a large receptive field. The experimental results on several publicly available datasets demonstrate the effectiveness of OSF-Logo. [ABSTRACT FROM AUTHOR]

As an essential subproblem in the face recognition field, occluded face recognition has received considerable attention in recent years. However, satisfactory recognition accuracy for occluded faces has not been achieved. Therefore, in order to solve this problem, this paper proposes a convolutional neural network that has multidimensional serial feature extraction modules for occluded faces and uses the deep learning method to improve the recognition rate. In order to improve the expression ability of the network, the method extracts features from two dimensions — space and channel — and built a multidimensional serial feature extraction module. By "multiscale" and "dependence" processing features, the generalization ability was significantly improved. Through the experimental test, the recognition accuracy of the multidimensional feature network (MFNet) reached 90.35%. This accuracy is 17.89% (72.36%) better than that of the traditional algorithm. Meanwhile, compared with other convolutional neural networks, this network has different degrees of improvement, which are 0.68% (ArcFace), 2.14% (Visual Geometry Group Face), and 4.83% (DeepID3). [ABSTRACT FROM AUTHOR]

• The network can directly learn end-to-end mapping between low-resolution images and high-resolution images, and it can use convolution and deconvolution cascade coding to eliminate noise. • Apply dilated convolution layer in the parts, so that small convolution kernel can obtain a large receptive field, which can improve the reconstruction effect and reduce the calculation amount. • Use residual learning method to quickly optimize the network. By adding multiple connections to the network, speed up the training speed and improve the performance. The existing image processing methods had aimed at the problems of blurred image reconstruction, large noise, and poor visual perception. The improved image super-resolution algorithm based on mixed deep convolutional networks is proposed in the paper. Firstly, the proposed method can shrink the low-resolution image to the specified size in upsampling phase. Secondly, it can extract features from low-resolution images. It sends the extracted initial features into the convolutional coding and decoding structure for image features. Thirdly, the feature extraction and calculation in high-dimensions are performed using dilated convolution in reconstruction layer. The high-resolution image had been reconstructed. The proposed method had been compared with state-of-arts on Set5, Set14, BSD100, and Urban100 datasets. The experimental results can show that the Peak Signal-to-Noise Ratio is increased by some ranges, and the Structural Similarity is increased by some effective percentage points. [Display omitted] [ABSTRACT FROM AUTHOR]

Regression-based methods (RBMs) have become a widely-used technique for feature extraction. However, most RBMs are only suitable for single-view data and fail to explore the consistency and complementarity information from multiple views. In this paper, we firstly propose a unified framework called locality cross-view regression (ULCR) to realize multi-view feature extraction. ULCR utilizes a regression loss function to explore the relationship between different views, meanwhile, preserving the manifold structure of samples. Then, under the ULCR framework, we propose a standard LCR (SLCR) which utilizes F-norm as the metric. SLCR is convenient for solving, but sensitive to the outliers. Therefore, a robust locality cross-view regression (RLCR) is proposed which uses L2,1-norm instead of F-norm in SLCR. The convergence analysis of the algorithm and the relationship between SLCR and RLCR are discussed. Experiment results on image datasets illustrate that the proposed methods develop better performance than other related methods. [ABSTRACT FROM AUTHOR]

Recent studies on three-dimensional (3D) point cloud data (PCD) simplification have played significant roles in computer-aided models for alleviating computational and storage burden. However, existing simplification methods are not suitable for the high-volume PCD with billions of point number, especially in construction industry. In this paper, a decomposed simplification method is developed to handle the high-volume buildings' PCD. Based on the divide-and-conquer philosophy, the new decomposed approach effectively reduces the memory usage. In the proposed approach, PCD is divided into several subsets according to the relationship of natural neighbor, and decomposed graph filtering with adaptive resampling rate is designed. It can be proved that the decomposed simplification results have no variance in the comparison with the optimal simplification of entire PCD. Verification experiments are conducted on different sizes of PCD, which indicate the effectiveness and feasibility of the proposed approach. • A decomposed method based on graph filtering is successfully designed for the high-volume PCD simplification problem. • A decomposed feature extraction algorithm is developed in high-volume PCD simplification algorithm. • Theoretical results guarantee the invariance of simplification in the proposed decomposed framework. [ABSTRACT FROM AUTHOR]

[Display omitted] With the development of smart sensors, large amount of operating data collected from a complex system as a high-speed train providing opportunities in efficient and effective fault detection and diagnosis (FDD). The data brings also challenges in the FDD modelling process, since the various signals may be redundant, useless and noisy for the FDD modelling of a specific sub-system. The data-driven methods suffer also from the curse of dimensionality. Feature dimension reduction can reduce the dimension of the monitoring dataset and eliminate the useless information. Different from the classical methods based on the correlation among variables, recent studies have shown that causality-based methods can make the FDD model more explanatory and robust. From the adjacency matrix of the causal network diagram, three unsupervised causality-based feature extraction methods for FDD in the braking system of a high-speed train are proposed in this paper. By constructing the causal network diagram among the raw monitoring feature variables through the causal discovery algorithm, the proposed methods extract informative features based on the causal adjacency matrix or the full causal adjacency matrix proposed in this work. These methods are adopted for fault detection with real dataset collected from the braking system in a high-speed train to verify their effectiveness. The experimental results show that the proposed causality-based feature extraction methods are effective and have certain advantages in comparison with the classical correlation-based methods. Especially, the feature extraction method based on the correlation matrix constructed from full causal adjacency matrix achieves better and stable results than the benchmark methods in the experiment. [ABSTRACT FROM AUTHOR]