Feature Vector Space Research Articles

Similarity Based Block Sparse Subset Selection for Video Summarization

Video summarization (VS) is generally formulated as a subset selection problem where a set of representative keyframes or key segments is selected from an entire video frame set. Though many sparse subset selection based VS algorithms have been proposed in the past decade, most of them adopt linear sparse formulation in the explicit feature vector space of video frames, and don’t consider the local or global relationships among frames. In this paper, we first extend the conventional sparse subset selection for VS into kernel block sparse subset selection (KBS3) to utilize the advantage of kernel sparse coding and introduce a local inter-frame relationship through packing of frame blocks. Going a step further, we propose a similarity based block sparse subset selection (SB2S3) model by applying a specially designed transformation matrix on the KBS3 model in order to introduce a kind of global inter-frame relationship through the similarity. Finally, a greedy pursuit based algorithm is devised for the proposed NP-hard model optimization. The proposed SB2S3 has the following advantages: 1) through the similarity between each frame and any other frame, the global relationship among all frames can be considered; 2) through block sparse coding, the local relationship of adjacent frames is further considered; and 3) it has a wider application, since features can derive similarity, but not vice versa. It is believed that the effect of modeling such global and local relationships among frames in this paper, is similar to that of modeling the long-range and short-range dependencies among frames in deep learning based methods. Experimental results on three benchmark datasets have demonstrated that the proposed approach is superior to not only other sparse subset selection based VS methods but also most unsupervised deep-learning based VS methods.

Hybrid feature CNN model for point cloud classification and segmentation

This study proposes a hybrid feature convolutional neural network (HFCNN) model for the complete description of three-dimensional (3D) point cloud features. The HFCNN confers sensitivity to the local, global, and single-point properties simultaneously by a feature vector space expansion. Wherein, a pointwise convolutional network sub-model realises the extraction of the local features by using a pointwise convolutional operator to process point cloud data directly. To consider the global properties of the point cloud, a central-point radiation model is constructed as an input of the feature layer in a non-network form. Meanwhile, the single-point behaviour is characterised by the solo point coordinate information in the network. Within the constructed solo-local-global feature space, i.e. the fusion of single point feature, local feature and global feature, the HFCNN model can handle 3D point cloud data with unstructured and unordered properties. The HFCNN can be directly applied to the point cloud classification and segmentation without the modification of the CNN structure and training procedure. The experimental results have shown the effectiveness of the proposed model in prediction of class labels and point-by-point labels.

Human context recognition with reduced feature space vectors for resource constraint gadgets

The aim of this paper is to determine the minimal features to train classification algorithms pertaining to context recognition using data collected from accelerometer sensor in smartphones. A detailed experimental evaluation of various time and frequency domain features on raw accelerometer sensor data collected from smartphones leads to the most influential minimal collection that aid in recognizing human context in resource constraint gadgets. To substantiate the study, four classifiers, namely, Logistic Regression, Support Vector Machine, Artificial Neural Network and Long Short Term Memory Recurrent Neural Networks are trained on six activities - Sitting, Standing and lying (sedentary activities), Walking, Walking Upstairs and Walking Downstairs (dynamic activities). The raw accelerometer values from UCI public dataset and the data collected from Android application is used to build classification model. Classifier performance on both datasets showed 90% accuracy with four features taken over each axes of 3-axis accelerometer.

A Clothing Retrieval Model Based on DenseNet and Multi-Similarity Loss

With the rapid development of deep learning in recent years, Deep Neural Network structure has been applied to all aspects of image algorithm, including image classification and recognition, target detection, image retrieval. Image retrieval is suffering from some challenges, such as crease, cover, background. Based on these challenges, using the local information of the image to replace the image is proposed which learns embedding vectors by local information to reduce noise interference. Inspired by the above discussion, a new deep learning model named IAPM (Integrated and Partial Model) is proposed in this paper. According to global information and local information, Joint training is carried out. For feature extraction, DenseNet is used to extract the feature vector. And the idea of transfer learning is applied to transform classification problems to retrieval problems. Meanwhile, the multi-task model is adopted to construct a framework, and Multi-Similarity Loss is used to pull the position of the embedded vector in feature space. Experimental results show its effectiveness.

BAT optimization based Retinal artery vein classification

The investigation of artery vein changes over time is considered to be the significant diagnosis process of retinal diseases like diabetic retinopathy. The diagnosis includes the characteristics analysis of artery vein vessels, changes in its tortuosity level and artery vein ratio; hence, it is important to classify the artery and vein in a better way. Computer-aided diagnosis requires the automated classification of retinal artery and vein for diagnosing the progression of diseases. In this paper, a supervised classification with Bat algorithm is proposed to discriminate the artery and vein vessels in the retinal fundus images. A novel feature vector space, including both additive colour space as well as luminous chromaticity model colour space, is constructed. BAT algorithm is applied to select the feature group which improve the classification accuracy and also to reduce the dimensionality of feature space. The proposed method is developed and analyzed using the publicly available databases DRIVE, IOSTAR and STARE.

(Invited) Domain Adaptation for the Semantic Segmentation of the Unmanned Surface Vehicle

Recently, deep learning has great development and is applied in various fields. But the performance of the network depends on a large amount of label data training. After training in a specific domain, a network model can have a good performance. But it may have a poor performance when transferred into a new domain. This paper addresses the visual semantic segmentation model transfer for the unmanned surface vehicle (USV) system. An adaptive cross-domain incremental learning method is proposed, which enables USV adaptive to navigate in unknown new environments. In our work, the knowledge of the trained segmentation network model and the feature information of the input image is used for the network incremental learning. First, the trained semantic segmentation network is used to classify each pixel of the input image. Then, pixels with high prediction probability are selected as seed points. At the same time, a convolution network for feature extraction is used to obtain the embedded feature space vector of each pixel. A mask centered on the seed point is generated based on the similarity between pixels in the feature space. The mask is labeled based on the segmentation network prediction. Multiple masks are fused to generate the label of the input image. Finally, the automatically generated labels are used to update the network to realize the adaptability of the network in the new environment

(Invited) Domain Adaptation for the Semantic Segmentation of the Unmanned Surface Vehicle

Recently, deep learning has great development and is applied in various fields. But the performance of the network depends on a large amount of label data training. After training in a specific domain, a network model can have a good performance. But it may have a poor performance when transferred into a new domain. This paper addresses the visual semantic segmentation model transfer for the unmanned surface vehicle (USV) system. An adaptive cross-domain incremental learning method is proposed, which enables USV adaptive to navigate in unknown new environments. In our work, the knowledge of the trained segmentation network model and the feature information of the input image is used for the network incremental learning. First, the trained semantic segmentation network is used to classify each pixel of the input image. Then, pixels with high prediction probability are selected as seed points. At the same time, a convolution network for feature extraction is used to obtain the embedded feature space vector of each pixel. A mask centered on the seed point is generated based on the similarity between pixels in the feature space. The mask is labeled based on the segmentation network prediction. Multiple masks are fused to generate the label of the input image. Finally, the automatically generated labels are used to update the network to realize the adaptability of the network in the new environment.

A principal component analysis-based feature dimensionality reduction scheme for content-based image retrieval system

In Content-Based Image Retrieval (CBIR) system, one approach of image representation is to employ combination of low-level visual features cascaded together into a flat vector. While this presents more descriptive information, it however poses serious challenges in terms of high dimensionality and high computational cost of feature extraction algorithms to deployment of CBIR on platforms (devices) with limited computational and storage resources. Hence, in this work a feature dimensionality reduction technique based on Principal Component Analysis (PCA) is implemented. Each image in a database is indexed using 174 dimensional feature vector comprising of 54-dimensional Colour Moments (CM54), 32-bin HSV-histogram (HIST32), 48-dimensional Gabor Wavelet (GW48) and 40-dimensional Wavelet Moments (MW40). The PCA scheme was incorporated into a CBIR system that utilized the entire feature vector space. The k-largest Eigenvalues that yielded a not more than 5% degradation in mean precision were retained for dimensionality reduction. Three image databases (DB10, DB20 and DB100) were used for testing. The result obtained showed that with 80% reduction in feature dimensions, tolerable loss of 3.45, 4.39 and 7.40% in mean precision value were achieved on DB10, DB20 and DB100.

The structural information filtered features (SIFF) potential: Maximizing information stored in machine-learning descriptors for materials prediction

Machine learning inspired potentials continue to improve the ability for predicting structures of materials. However, many challenges still exist, particularly when calculating structures of disordered systems. These challenges are primarily due to the rapidly increasing dimensionality of the feature-vector space which in most machine-learning algorithms is dependent on the size of the structure. In this article, we present a feature-engineered approach that establishes a set of principles for representing potentials of physical structures (crystals, molecules, and clusters) in a feature space rather than a physically motivated space. Our goal in this work is to define guiding principles that optimize information storage of the physical parameters within the feature representations. In this manner, we focus on keeping the dimensionality of the feature space independent of the number of atoms in the structure. Our Structural Information Filtered Features (SIFF) potential represents structures by utilizing a feature vector of low-correlated descriptors, which correspondingly maximizes information within the descriptor. We present results of our SIFF potential on datasets composed of disordered (carbon and carbon–oxygen) clusters, molecules with C7O2H2 stoichiometry in the GDB9-14B dataset, and crystal structures of the form (AlxGayInz)2O3 as proposed in the NOMAD Kaggle competition. Our potential's performance is at least comparable, sometimes significantly more accurate, and often more efficient than other well-known machine-learning potentials for structure prediction. However, primarily, we offer a different perspective on how researchers should consider opportunities in maximizing information storage for features.

A Graph-Based Feature Generation Approach in Android Malware Detection with Machine Learning Techniques

An explosive spread of Android malware causes a serious concern for Android application security. One of the solutions to detecting malicious payloads sneaking in an application is to treat the detection as a binary classification problem, which can be effectively tackled with traditional machine learning techniques. The key factors in detecting Android malware with machine learning techniques are feature selection and generation. Most of the existing approaches select and generate features without fully examining the structures of programs, and thus the important semantic information associated with these features is lost, consequently resulting in a low accuracy rate in detection. To address this issue, we propose a new feature generation approach for Android applications, which takes components and program structures into consideration and extracts features in a graph-based and semantics-rich style. This approach highlights two major distinguishing aspects: the context-based feature selection and graph-based feature generation. We abstract an Android application as a collection of reduced iCFGs (interprocedural control flow graphs) and extract original features from these graphs. Combining the original features and their contexts together, we generate new features which hold richer semantic information than the original ones. By embedding the features into a feature vector space, we can use machine learning techniques to train a malware detector. The experiment results show that this approach achieves an accuracy rate of 95.4% and a recall rate of 96.5%, which prove the effectiveness and advantages of our approach.

RF-PseU: A Random Forest Predictor for RNA Pseudouridine Sites.

One of the ubiquitous chemical modifications in RNA, pseudouridine modification is crucial for various cellular biological and physiological processes. To gain more insight into the functional mechanisms involved, it is of fundamental importance to precisely identify pseudouridine sites in RNA. Several useful machine learning approaches have become available recently, with the increasing progress of next-generation sequencing technology; however, existing methods cannot predict sites with high accuracy. Thus, a more accurate predictor is required. In this study, a random forest-based predictor named RF-PseU is proposed for prediction of pseudouridylation sites. To optimize feature representation and obtain a better model, the light gradient boosting machine algorithm and incremental feature selection strategy were used to select the optimum feature space vector for training the random forest model RF-PseU. Compared with previous state-of-the-art predictors, the results on the same benchmark data sets of three species demonstrate that RF-PseU performs better overall. The integrated average leave-one-out cross-validation and independent testing accuracy scores were 71.4% and 74.7%, respectively, representing increments of 3.63% and 4.77% versus the best existing predictor. Moreover, the final RF-PseU model for prediction was built on leave-one-out cross-validation and provides a reliable and robust tool for identifying pseudouridine sites. A web server with a user-friendly interface is accessible at http://148.70.81.170:10228/rfpseu.

Multi-scale affined-HOF and dimension selection for view-unconstrained action recognition

In this paper an action recognition method that can adaptively handle the problems of variations in camera viewpoint is introduced. Our contribution is three-fold. First, a space-sampling algorithm based on affine transform in multiple scales is proposed to yield a series of different viewpoints from a single one. A histogram of dense optical flow is then extracted over each fixed-size patch for a given generated viewpoint as a local feature descriptor. Second, a dimension selection procedure is also proposed to retain only the dimensions that have distinctive information and discard the unnecessary ones in the feature vector space. Third, to adapt to a situation in which video data in multiple viewpoints are used for training; an extended method with a voting algorithm is also introduced to increase the recognition accuracy. By conducting experiments using both simulated and realistic datasets (http://www.aislab.org/index.php/en/mvar-datasets), the proposed method is validated. The method is found to be accurate and capable of maintaining its accuracy under a wide range of viewpoint changes. In addition, the method is less sensitive to variations in subject scale, subject position, action speed, partial occlusion, and background. The method is also validated by comparing with state-of-the-art view-invariant action recognition methods using well-known i3DPost and MuHAVi public datasets.

PTPG: A Poisson Triangular Pattern Generator for Tokens on Tangible Surfaces

We present PTPG (Poisson triangular pattern generator), a method that generates and optimizes triangular patterns for tokens on tangible surfaces. PTPG uses spatial information of markers to form triangular patterns and processes a template-matching algorithm to identify patterns. We introduce Poisson disk sampling to search patterns in the feature vector space and set sampling constraints to obtain patterns that meet the requirements of size, shape, and pattern distance. The patterns optimized by a shape constraint address the marker-loss problem caused by a common design of capacitive tokens, and therefore improve the recognition rate. We manufacture ten sets of prototype tokens based on the patterns generated by Poisson sampling and test these tokens on different devices. Experimental results show that our method provides more unique patterns than other heuristic methods, and the patterns are applicable to different tangible surfaces with high recognition rates.

Recognition method of single trial motor imagery electroencephalogram signal based on sparse common spatial pattern and Fisher discriminant analysis

This paper aims to realize the decoding of single trial motor imagery electroencephalogram (EEG) signal by extracting and classifying the optimized features of EEG signal. In the classification and recognition of multi-channel EEG signals, there is often a lack of effective feature selection strategies in the selection of the data of each channel and the dimension of spatial filters. In view of this problem, a method combining sparse idea and greedy search (GS) was proposed to improve the feature extraction of common spatial pattern (CSP). The improved common spatial pattern could effectively overcome the problem of repeated selection of feature patterns in the feature vector space extracted by the traditional method, and make the extracted features have more obvious characteristic differences. Then the extracted features were classified by Fisher linear discriminant analysis (FLDA). The experimental results showed that the classification accuracy obtained by proposed method was 19% higher on average than that of traditional common spatial pattern. And high classification accuracy could be obtained by selecting feature set with small size. The research results obtained in the feature extraction of EEG signals lay the foundation for the realization of motor imagery EEG decoding.

Training data independent image registration using generative adversarial networks and domain adaptation

Medical image registration is an important task in automated analysis of multimodal images and temporal data involving multiple patient visits. Conventional approaches, although useful for different image types, are time consuming. Of late, deep learning (DL) based image registration methods have been proposed that outperform traditional methods in terms of accuracy and time. However, DL based methods are heavily dependent on training data and do not generalize well when presented with images of different scanners or anatomies. We present a DL based approach that can perform medical image registration of one image type despite being trained with images of a different type. This is achieved by unsupervised domain adaptation in the registration process and allows for easier application to different datasets without extensive retraining. To achieve our objective we train a network that transforms the given input image pair to a latent feature space vector using autoencoders. The resultant encoded feature space is used to generate the registered images with the help of generative adversarial networks (GANs). This feature transformation ensures greater invariance to the input image type. Experiments on chest X-ray, retinal and brain MR images show that our method, trained on one dataset gives better registration performance for other datasets, outperforming conventional methods that do not incorporate domain adaptation.

Spatial Ensemble Learning for Heterogeneous Geographic Data with Class Ambiguity

Class ambiguity refers to the phenomenon whereby similar features correspond to different classes at different locations. Given heterogeneous geographic data with class ambiguity, the spatial ensemble learning (SEL) problem aims to find a decomposition of the geographic area into disjoint zones such that class ambiguity is minimized and a local classifier can be learned in each zone. The problem is important for applications such as land cover mapping from heterogeneous earth observation data with spectral confusion. However, the problem is challenging due to its high computational cost. Related work in ensemble learning either assumes an identical sample distribution (e.g., bagging, boosting, random forest) or decomposes multi-modular input data in the feature vector space (e.g., mixture of experts, multimodal ensemble) and thus cannot effectively minimize class ambiguity. In contrast, we propose a spatial ensemble framework that explicitly partitions input data in geographic space. Our approach first preprocesses data into homogeneous spatial patches and uses a greedy heuristic to allocate pairs of patches with high class ambiguity into different zones. We further extend our spatial ensemble learning framework with spatial dependency between nearby zones based on the spatial autocorrelation effect. Both theoretical analysis and experimental evaluations on two real world wetland mapping datasets show the feasibility of the proposed approach.

Age-invariant face recognition using multiple descriptors along with modified dimensionality reduction approach

Face recognition on the basis of age variation is a significant yet challenging issue. One among the effective methods to age-invariant face recognition is to form a face aging model that can be utilized to recompense for the aging process in face matching or age assessment. The periocular region of a face is the most age-invariant facial region, to abstract discriminative local features that are distinct for every subject. Feature vector space can be reduced by utilizing the features only from the periocular region. So, in this article, features are only extracted from the periocular region of a face. For feature extraction, multiple descriptors such as Scale Invariant Feature Transform (SIFT) and then Speeded Up Robust Features (SURF). As the extracted features vector has high dimensionality, it is decreased to the low dimensionality using the Enhanced Principal Components Analysis (EPCA) method. Finally, these extracted features are given as input to the Artificial Neural Network (ANN) based classifier which performs to recognize the face of the input image. Our projected method is applied and tested by Matlab for FG-NET face aging dataset Simulation results show that the performance of the proposed approach outperforms that of the existing age-invariant face recognition schemes in terms of accuracy, complexity and false recognition ratio.

Visual Monitoring of Industrial Operation States Based on Kernel Fisher Vector and Self-organizing Map Networks

As industrial process becomes increasingly complicated and the correlation between industrial process variables tends to exhibit strong nonlinear characteristics, how to effectively and visually monitor industrial operation states is challenging. A method based on kernel Fisher vector and self-organizing map networks (KFV-SOM) is proposed to improve the visualization of process monitoring. In KFV-SOM, kernel Fisher discriminant analysis is first employed to map data into high-dimensional space by using a nonlinear function, and the optimal Fisher feature vector, which can represent industrial operation states fittingly, is extracted. That is, the normal state and different kinds of faults can be distinguished well in the Fisher feature vector space. The topological structure of the Fisher feature vector space is then visualized intuitively on the two-dimensional output map of self-organizing map (SOM) with the Fisher feature vector as the input of the SOM network. Thus, the KFV-SOM can effectively realize the visualization of monitoring. Continuous stirred tank reactor process is applied to illustrate the capability of KFV-SOM. Result shows that KFV-SOM can effectively visualize monitoring, and it is better in showing the operation states of normal state and different kinds of faults on the output map of the SOM network than SOM, SOM integrated with principal component analysis, SOM integrated with correlative component analysis, SOM integrated with Fisher discriminant analysis, and SOM integrated with canonical variable analysis.

Fault Diagnosis for Rolling Element Bearings Based on Feature Space Reconstruction and Multiscale Permutation Entropy.

Aimed at distinguishing different fault categories of severity of rolling bearings, a novel method based on feature space reconstruction and multiscale permutation entropy is proposed in the study. Firstly, the ensemble empirical mode decomposition algorithm (EEMD) was employed to adaptively decompose the vibration signal into multiple intrinsic mode functions (IMFs), and the representative IMFs which contained rich fault information were selected to reconstruct a feature vector space. Secondly, the multiscale permutation entropy (MPE) was used to calculate the complexity of reconstructed feature space. Finally, the value of multiscale permutation entropy was presented to a support vector machine for fault classification. The proposed diagnostic algorithm was applied to three groups of rolling bearing experiments. The experimental results indicate that the proposed method has better classification performance and robustness than other traditional methods.

Automatic Land Cover Reconstruction From Historical Aerial Images: An Evaluation of Features Extraction and Classification Algorithms.

The land cover reconstruction from monochromatic historical aerial images is a challenging task that has recently attracted an increasing interest from the scientific community with the proliferation of large-scale epidemiological studies involving retrospective analysis of spatial patterns. However, the efforts made by the computer vision community in remote-sensing applications are mostly focused on prospective approaches through the analysis of high-resolution multi-spectral data acquired by the advanced spatial programs. Hence, four contributions are proposed in this paper. They aim at providing a comparison basis for the future development of computer vision algorithms applied to the automation of the land cover reconstruction from monochromatic historical aerial images. First, a new multi-scale multi-date dataset composed of 4.9 million non-overlapping annotated patches of the France territory between 1970 and 1990 has been created with the help of geography experts. This dataset has been named HistAerial. Second, an extensive comparison study of the state-of-the-art texture features extraction and classification algorithms, including deep convolutional neural networks (DCNNs), has been performed. It is presented in the form of an evaluation. Third, a novel low-dimensional local texture filter named rotated-corner local binary pattern (R-CRLBP) is presented as a simplification of the binary gradient contours filter through the use of an orthogonal combination representation. Finally, a novel combination of low-dimensional texture descriptors, including the R-CRLBP filter, is introduced as a light combination of local binary patterns (LCoLBPs). The LCoLBP filter achieved state-of-the-art results on the HistAerial dataset while conserving a relatively low-dimensional feature vector space compared with the DCNN approaches (17 times shorter).