Classes In Feature Space Research Articles

Rebalancing network with knowledge stability for class incremental learning

Class incremental learning (CIL) has been proposed to solve the problem of learning to classify new classes while maintaining the performance on old classes. A typical strategy is to update the old classification model with entire new class data and a few old exemplars, which face serious performance degradation on old classes. The main reasons come down to class imbalance between old and new classes along with catastrophic forgetting towards old classes. Most existing CIL methods have proposed solving the above two issues in classification space, ignoring their adverse effects of overlapping between old and new classes and confusion among old classes in feature space. In this paper, we propose a rebalancing network with knowledge stability (RNKS), aiming to adequately retain the model performance on old classes in CIL by solving class imbalance and catastrophic forgetting in feature and classification space simultaneously. In detail, the proposed RNKS mainly consists of multi-proxies rebalancing (MPR) and hybrid knowledge distillation (HKD). MPR, focusing on class imbalance, employs multi-proxies metric learning to decrease the feature overlapping between old and new classes, together with balanced data sampling to correct the skewed decision boundary. HKD, coping with catastrophic forgetting, encourages the updated model to reproduce identical feature topologies and predictions of old classes as the old model via feature relation-based and response-based distillations. Experiments on CIFAR-100 and ILSVRC datasets demonstrate the effectiveness of this work against the state-of-the-art approaches.

Gabor wavelet and neural network face detection

One of the most difficult tasks in image processing is facial area detection. This study introduces a new face detection method. To improve detection rates, the system incorporates two facial detection algorithms. Gabor wavelets and neural networks are the two algorithms. Convolutional face images undergo initial transformation using Gabor wavelets, with 8 orientations and 5 scales chosen to extract the grey characteristics of the facial region. When added to the original photos, these 40 Gabor wavelets reveal the full extent of the response. We use a second feedforward neural network specifically designed for facial detection. The neural network is trained by backpropagation using the training set of faces and non-faces. Our experiments show that the suggested Gabor wavelet faces, when combined with the neural network feature space classifier, provide very respectable results. Comparing our proposed system to other face detection systems reveals that it performs better in terms of detection and false negative rates.

Distance metric-based learning for long-tail object detection

Despite the recent success of general object detection, almost all models perform unsatisfactorily on long-tail datasets. The main cause of performance degradation is the imbalance in the number of positive samples between categories. The traditional approaches can lead to distortion of the classification feature space, which in turn can seriously affect the classification ability of the network. To address the above issues, we propose a novel distance metric-based learning approach for long-tail object detection (LTDL) in this paper. Specifically, we directly use the feature space as the optimization target, thus allowing clearer decision boundaries between classes. In order to optimize the decision boundary, we adjust the intra-class and inter-class distances by Margin Module (MAM). Meanwhile, to further exploit the information provided by the dataset, we introduce supervised information of labels for distance weighting using the Semantic Module (SEM). In addition, to protect the learning of tail samples and optimize the classifier, we propose a Distance-based Equilibrium Loss (DEL). Extensive experiments conducted on the LVIS benchmark have demonstrated the strength of our proposed approach. The experimental results show that our method improves the baseline by 2.9% AP. And our best model can outperform almost all other representative methods.

Youtube Comment Feature Selection And Classification Using Fused Machine Learning

The exponential rise of internet platforms, notably YouTube, has resulted in a massive volume of user-generated material, including video comments. Understanding audience input and enhancing user experience require analyzing and forecasting the mood of YouTube comments. This work provides a comprehensive method for YouTube comment prediction and sentiment classification that combines feature selection using Recursive Feature Elimination (RFE), Elastic Net Random Forest with Logistic Regression (RF with LR), and Principal Component Analysis (PCA). The first stage is to choose the most informative features from a given dataset using RFE, a common approach for doing so. RFE aids in the elimination of unnecessary or redundant features, resulting in enhanced model performance and decreased computing complexity. The Elastic Net Random Forest with Logistic Regression (RF with LR) technique is then used to construct a robust sentiment classification model. Elastic Net regularization combines the advantages of both L1 (Lasso) and L2 (Ridge) regularization, allowing for improved feature selection and management of multicollinearity concerns. By integrating many decision trees, the Random Forest ensemble approach improves the model's predictive potential even more. We employ Principal Component Analysis (PCA) to increase the classification model's effectiveness and solve possible difficulties created by high-dimensional data. PCA decreases the dataset's complexity while retaining its fundamental qualities, resulting in a more manageable and efficient feature space for classification. Finally, we compare the performance of three prominent classifiers on the preprocessed dataset: Linear Support Vector Machine (LSVM), Gaussian Naive Bayes (GNB), Logistic Regression (LR), and Decision Tree (DT). We can select the best-performing model for YouTube comment categorization by comparing these classifiers.

Webly-supervised semantic segmentation via curriculum learning

In this paper, we propose a weakly supervised semantic segmentation method by directly learning from web images, which are crawled from the Internet by using text queries, without any explicit user annotation or even data filtering. With the goal of handling the massive amount of noisy labels in web images, we design a three-stage approach for weakly-supervised semantic segmentation based on curriculum learning. We first generate pixel-level masks for the training images via a popular weakly-supervised semantic segmentation framework. Then, we consider the noise of the web data in two ways. At the image-level, the complexity of data is measured using its distribution density in a classification feature space. At the pixel-level, the complexity of the mask is evaluated by exploiting the relationship between the saliency map and those segmented images in an unsupervised manner. The key insight to this design is that, common and simple object patterns in images should be salient with both the saliency detector and weakly supervised DCNNs, where they should be sparse with high regional consistency between them. This allows for an efficient implementation of curriculum learning from noisy web images. Experiments on the popular PASCAL VOC 2012 benchmark show that we achieve very competitive performance with scores of 64.0% mIoU using our pure web dataset, which contains noisy, single-label images. We further improve the performance to 69.7% mIoU by using the CurriculumWebSegNet fine-tuned on the PASCAL VOC dataset, which has more precise multi-label supervision.

Imitating the oracle: Towards calibrated model for class incremental learning

Class-incremental learning (CIL) aims to recognize classes that emerged in different phases. The joint-training (JT), which trains the model jointly with all classes, is often considered as the upper bound of CIL. In this paper, we thoroughly analyze the difference between CIL and JT in feature space and weight space. Motivated by the comparative analysis, we propose two types of calibration: feature calibration and weight calibration to imitate the oracle (ItO), i.e., JT. Specifically, on the one hand, feature calibration introduces deviation compensation to maintain the class decision boundary of old classes in feature space. On the other hand, weight calibration leverages forgetting-aware weight perturbation to increase transferability and reduce forgetting in parameter space. With those two calibration strategies, the model is forced to imitate the properties of joint-training at each incremental learning stage, thus yielding better CIL performance. Our ItO is a plug-and-play method and can be implemented into existing methods easily. Extensive experiments on several benchmark datasets demonstrate that ItO can significantly and consistently improve the performance of existing state-of-the-art methods. Our code is publicly available at https://github.com/Impression2805/ItO4CIL.

Classification and visual explanation for COVID-19 pneumonia from CT images using triple learning

This study presents a novel framework for classifying and visualizing pneumonia induced by COVID-19 from CT images. Although many image classification methods using deep learning have been proposed, in the case of medical image fields, standard classification methods are unable to be used in some cases because the medical images that belong to the same category vary depending on the progression of the symptoms and the size of the inflamed area. In addition, it is essential that the models used be transparent and explainable, allowing health care providers to trust the models and avoid mistakes. In this study, we propose a classification method using contrastive learning and an attention mechanism. Contrastive learning is able to close the distance for images of the same category and generate a better feature space for classification. An attention mechanism is able to emphasize an important area in the image and visualize the location related to classification. Through experiments conducted on two-types of classification using a three-fold cross validation, we confirmed that the classification accuracy was significantly improved; in addition, a detailed visual explanation was achieved comparison with conventional methods.

Ice reconnaissance data processing under low quality source images

A practically effective solution to the problem of automated processing of ice reconnaissance data in high latitudes is proposed. The intermediate result of ice reconnaissance is huge aerial survey data set consisting of images of low quality; this is a consequence of the difficult conditions of aerial survey in high latitudes. The goal of the study is to create a high-level method that can either efficiently process this pre-collected data set or perform real-time processing of similar images while ensuring high reliability in solving the problem of recognizing ice class distribution on the water surface with minimal computing resources. In particular, the problem of automatic classification of ice-floe size distribution (FSD) type for a three-class model based on aerial survey data is solved. The practically important case of low-quality images is considered, a common situation for the meteorological conditions of the Far North. The proposed approach is based on the use of machine learning methods, in particular on the well-known multi-class SVM (Support Vector Machine), which is extremely undemanding to computing resources and therefore can be implemented even by the onboard computer of an ice reconnaissance UAV. From the input images of low quality some numerical characteristics of the image are calculated which informatively characterize the image. These characteristics (features) are invariant to scaling, rotation and illumination as well as have a much smaller dimensionality than the original image. The main idea underlying the proposed method is to form an original set of features which are implemented in the original feature space. These features characterize large fragments of the analyzed image and are “stable”, in contrast to the features that characterize small details. A new method of FSD type classification based on the processing of aerial survey data by using machine learning methods, which is sufficiently effective for processing low-quality images, has been proposed. Also, the original feature space for classification was proposed which ensured high practical efficiency of this method. The method has shown high efficiency when it is tested on a data set composed of low-quality real images (high blurriness, vagueness, presence of meteorological noises). The developed algorithm can be used for express analysis of ice reconnaissance data, including an ice reconnaissance UAV on-board software component.

Semi-supervised semantic segmentation with cross teacher training

Convolutional neural networks can achieve remarkable performance in semantic segmentation tasks. However, such neural network approaches heavily rely on costly pixel-level annotation. Semi-supervised learning is a promising resolution to tackle this issue, but its performance still far falls behind the fully supervised counterpart. This work proposes a cross-teacher training framework with three modules that significantly improves traditional semi-supervised learning approaches. The core is a cross-teacher module, which could simultaneously reduce the coupling among peer networks and the error accumulation between teacher and student networks. In addition, we propose two complementary contrastive learning modules. The high-level module can transfer high-quality knowledge from labeled data to unlabeled ones and promote separation between classes in feature space. The low-level module can encourage low-quality features learning from the high-quality features among peer networks. In experiments, the cross-teacher module significantly improves the performance of traditional student–teacher approaches, and our framework outperforms state-of-the-art methods on benchmark datasets.

Fault Diagnosis Method for Rolling Bearing Based on Sparse Principal Subspace Discriminant Analysis

Rolling bearings are omnipresent parts in industrial fields. To comprehensively reflect the status of rolling bearing and improve the classification accuracy, fusion information is widely used in various studies, which may result in high dimensionality, redundancy information of dataset, and time consumption. Thus, it is of crucial significance in extracting optimal features from high-dimensional and redundant feature space for classification. In this study, a fault diagnosis of rolling bearings model based on sparse principal subspace discriminant analysis is proposed. It extracts sparse discrimination information, meanwhile preserving the main energy of original dataset, and the sparse regularization term and sparse error term constrained by l2,1-norm are introduced to improve the performance of feature extraction and the robustness to noise and outliers. The multi-domain feature space involved a time domain, frequency domain, and time-frequency domain is first derived from the original vibration signals. Then, the intrinsic geometric features extracted by sparse principal subspace discriminant analysis are fed into a support vector machine classifier to recognize different operating conditions of bearings. The experimental results demonstrated that the feasibility and effectiveness of the proposed fault diagnosis model based on a sparse principal subspace discriminant analysis algorithm can achieve higher recognition accuracy than fisher discriminant analysis and its extensions, and it is relatively insensitive to the impact of noise and outliers owing to the sparse property.

Improving Facial Emotion Recognition Using Residual Autoencoder Coupled Affinity Based Overlapping Reduction

Emotion recognition using facial images has been a challenging task in computer vision. Recent advancements in deep learning has helped in achieving better results. Studies have pointed out that multiple facial expressions may present in facial images of a particular type of emotion. Thus, facial images of a category of emotion may have similarity to other categories of facial images, leading towards overlapping of classes in feature space. The problem of class overlapping has been studied primarily in the context of imbalanced classes. Few studies have considered imbalanced facial emotion recognition. However, to the authors’ best knowledge, no study has been found on the effects of overlapped classes on emotion recognition. Motivated by this, in the current study, an affinity-based overlap reduction technique (AFORET) has been proposed to deal with the overlapped class problem in facial emotion recognition. Firstly, a residual variational autoencoder (RVA) model has been used to transform the facial images to a latent vector form. Next, the proposed AFORET method has been applied on these overlapped latent vectors to reduce the overlapping between classes. The proposed method has been validated by training and testing various well known classifiers and comparing their performance in terms of a well known set of performance indicators. In addition, the proposed AFORET method is compared with already existing overlap reduction techniques, such as the OSM, ν-SVM, and NBU methods. Experimental results have shown that the proposed AFORET algorithm, when used with the RVA model, boosts classifier performance to a greater extent in predicting human emotion using facial images.

Contrastive Learning for Mitochondria Segmentation.

Mitochondria segmentation in electron microscopy images is essential in neuroscience. However, due to the image degradation during the imaging process, the large variety of mitochondrial structures, as well as the presence of noise, artifacts and other sub-cellular structures, mitochondria segmentation is very challenging. In this paper, we propose a novel and effective contrastive learning framework to learn a better feature representation from hard examples to improve segmentation. Specifically, we adopt a point sampling strategy to pick out representative pixels from hard examples in the training phase. Based on these sampled pixels, we introduce a pixel-wise label-based contrastive loss which consists of a similarity loss term and a consistency loss term. The similarity term can increase the similarity of pixels from the same class and the separability of pixels from different classes in feature space, while the consistency term is able to enhance the sensitivity of the 3D model to changes in image content from frame to frame. We demonstrate the effectiveness of our method on MitoEM dataset as well as FIB-SEM dataset and show better or on par with state-of-the-art results.

Multiclass Anomaly Detection in Flight Data Using Semi-Supervised Explainable Deep Learning Model

The identification of precursors to safety incidents in aviation data is a crucial task, yet extremely challenging. The main approach in practice leverages domain expertise to define expected tolerances in system behavior and flags exceedances from such safety margins. However, this approach is incapable of identifying unknown risks and vulnerabilities. Various machine-learning approaches have been investigated and deployed to identify anomalies, with the great challenge of procuring enough labeled data to achieve reliable and accurate performance. This paper presents an explainable deep semi-supervised model for anomaly detection in aviation, building upon recent advancements described in the machine-learning literature. The proposed model combines feature engineering and classification in feature space, while leveraging all available data (labeled and unlabeled). Our approach is validated with case studies of anomaly detection during the takeoff and landing phases of commercial aircraft. Our model outperforms the state-of-the-art supervised anomaly-detection model, reaching significantly higher accuracy and fewer false alarms, even if only small proportion of data in the training set is labeled.

Early Identification of Seed Maize and Common Maize Production Fields Using Sentinel-2 Images

Accurate and timely access to the production area of crop seeds allows the seed market and secure seed supply to be monitored. Seed maize and common maize production fields typically share similar phenological development profiles with differences in the planting patterns, which makes it challenging to separate these fields from decametric-resolution satellite images. In this research, we proposed a method to identify seed maize production fields as early as possible in the growing season using a time series of remote sensing images in the Liangzhou district of Gansu province, China. We collected Sentinel-2 and GaoFen-1 (GF-1) images captured from March to September. The feature space for classification consists of four original bands, namely red, green, blue, and near-infrared (nir), and eight vegetation indexes. We analyzed the timeliness of seed maize identification using Sentinel-2 time series of different time spans and identified the earliest time frame for reasonable classification accuracy. Then, the earliest time series that met the requirements of regulatory accuracy were compared and analyzed. Four machine/deep learning algorithms were tested, including K-nearest neighbor (KNN), support vector classification (SVC), random forest (RF), and long short-term memory (LSTM). The results showed that using Sentinel-2 images from March to June, the RF and LSTM algorithms achieve over 88% accuracy, with the LSTM performing the best (90%). In contrast, the accuracy of KNN and SVC was between 82% and 86%. At the end of June, seed maize mapping can be carried out in the experimental area, and the precision can meet the basic requirements of monitoring for the seed industry. The classification using GF-1 images were less accurate and reliable; the accuracy was 85% using images from March to June. To achieve near real-time identification of seed maize fields early in the growing season, we adopted an automated sample generation approach for the current season using only historical samples based on clustering analysis. The classification accuracy using new samples extracted from historical mapping reached 74% by the end of the season (September) and 63% by the end of July. This research provides important insights into the classification of crop fields cultivated with the same crop but different planting patterns using remote sensing images. The approach proposed by this study enables near-real time identification of seed maize production fields within the growing season, which could effectively support large-scale monitoring of the seed supply industry.

Joint Label Prediction Based Semi-Supervised Adaptive Concept Factorization for Robust Data Representation

Constrained Concept Factorization (CCF) yields the enhanced representation ability over CF by incorporating label information as additional constraints, but it cannot classify and group unlabeled data appropriately. Minimizing the difference between the original data and its reconstruction directly can enable CCF to model a small noisy perturbation, but is not robust to gross sparse errors. Besides, CCF cannot preserve the manifold structures in new representation space explicitly, especially in an adaptive manner. In this paper, we propose a joint label prediction based Robust Semi-Supervised Adaptive Concept Factorization (RS2ACF) framework. To obtain robust representation, RS2ACF relaxes the factorization to make it simultaneously stable to small entrywise noise and robust to sparse errors. To enrich prior knowledge to enhance the discrimination, RS2ACF clearly uses class information of labeled data and more importantly propagates it to unlabeled data by jointly learning an explicit label indicator for unlabeled data. By the label indicator, RS2ACF can ensure the unlabeled data of the same predicted label to be mapped into the same class in feature space. Besides, RS2ACF incorporates the joint neighborhood reconstruction error over the new representations and predicted labels of both labeled and unlabeled data, so the manifold structures can be preserved explicitly and adaptively in the representation space and label space at the same time. Owing to the adaptive manner, the tricky process of determining the neighborhood size or kernel width can be avoided. Extensive results on public databases verify that our RS2ACF can deliver state-of-the-art data representation, compared with other related methods.

Differentiating Peripherally-Located Small Cell Lung Cancer From Non-small Cell Lung Cancer Using a CT Radiomic Approach.

Lung cancer can be classified into two main categories: small cell lung cancer (SCLC) and non-small cell lung cancer (NSCLC), which are different in treatment strategy and survival probability. The lung CT images of SCLC and NSCLC are similar such that their subtle differences are hardly visually discernible by the human eye through conventional imaging evaluation. We hypothesize that SCLC/NSCLC differentiation could be achieved via computerized image feature analysis and classification in feature space, as termed a radiomic model. The purpose of this study was to use CT radiomics to differentiate SCLC from NSCLC adenocarcinoma. Patients with primary lung cancer, either SCLC or NSCLC adenocarcinoma, were retrospectively identified. The post-diagnosis pre-treatment lung CT images were used to segment the lung cancers. Radiomic features were extracted from histogram-based statistics, textural analysis of tumor images and their wavelet transforms. A minimal-redundancy-maximal-relevance method was used for feature selection. The predictive model was constructed with a multilayer artificial neural network. The performance of the SCLC/NSCLC adenocarcinoma classifier was evaluated by the area under the receiver operating characteristic curve (AUC). Our study cohort consisted of 69 primary lung cancer patients with SCLC (n = 35; age mean ± SD = 66.91± 9.75 years), and NSCLC adenocarcinoma (n = 34; age mean ± SD = 58.55 ± 11.94 years). The SCLC group had more male patients and smokers than the NSCLC group (P < 0.05). Our SCLC/NSCLC classifier achieved an overall performance of AUC of 0.93 (95% confidence interval = [0.85, 0.97]), sensitivity = 0.85, and specificity = 0.85). Adding clinical data such as smoking history could improve the performance slightly. The top ranking radiomic features were mostly textural features. Our results showed that CT radiomics could quantitatively represent tumor heterogeneity and therefore could be used to differentiate primary lung cancer subtypes with satisfying results. CT image processing with the wavelet transformation technique enhanced the radiomic features for SCLC/NSCLC classification. Our pilot study should motivate further investigation of radiomics as a non-invasive approach for early diagnosis and treatment of lung cancer.

SVDTWDD Method for High Correct Recognition Rate Classifier With Appropriate Rejection Recognition Regions

At present, regions of the same class determined by Support Vector Machines (SVM) classifier, Support Vector Domain Description (SVDD) classifier and Deep Learning (DL) classifier may occupy regions of other classes or unknown classes in feature space. There exists a risk that samples of other classes or unknown classes are wrongly classified as a known class. In this paper, the Support Vector Domain Tightly Wrapping Description Design (SVDTWDD) method with appropriate rejection regions and the corresponding incremental learning algorithm are proposed to overcome the above problem. The main work includes: (1) We develop a construction algorithm of the tightly wrapping set for the homogeneous feature set; (2) Based on the homogeneous feature set and tightly wrapping set, a novel algorithm is presented for obtaining the tightly wrapping surface of the homogeneous feature region; (3) The method for constructing all the public regions outside of the tightly wrapping surface and the intersections of wrapping regions in two different tightly wrapping surfaces, as the rejection region of the classifier; (4) An incremental algorithm is also presented based on the SVD-TWDD method. The experimental results with UCI data sets show that the correct recognition rate of our proposed method is nearly100% even if with a low rejection rate.

Siamese Reconstruction Network: Accurate Image Reconstruction from Human Brain Activity by Learning to Compare

Decoding human brain activities, especially reconstructing human visual stimuli via functional magnetic resonance imaging (fMRI), has gained increasing attention in recent years. However, the high dimensionality and small quantity of fMRI data impose restrictions on satisfactory reconstruction, especially for the reconstruction method with deep learning requiring huge amounts of labelled samples. When compared with the deep learning method, humans can recognize a new image because our human visual system is naturally capable of extracting features from any object and comparing them. Inspired by this visual mechanism, we introduced the mechanism of comparison into deep learning method to realize better visual reconstruction by making full use of each sample and the relationship of the sample pair by learning to compare. In this way, we proposed a Siamese reconstruction network (SRN) method. By using the SRN, we improved upon the satisfying results on two fMRI recording datasets, providing 72.5% accuracy on the digit dataset and 44.6% accuracy on the character dataset. Essentially, this manner can increase the training data about from n samples to 2n sample pairs, which takes full advantage of the limited quantity of training samples. The SRN learns to converge sample pairs of the same class or disperse sample pairs of different class in feature space.

Online Learning from Data Streams with Varying Feature Spaces

We study the problem of online learning with varying feature spaces. The problem is challenging because, unlike traditional online learning problems, varying feature spaces can introduce new features or stop having some features without following a pattern. Other existing methods such as online streaming feature selection (Wu et al. 2013), online learning from trapezoidal data streams (Zhang et al. 2016), and learning with feature evolvable streams (Hou, Zhang, and Zhou 2017) are not capable to learn from arbitrarily varying feature spaces because they make assumptions about the feature space dynamics. In this paper, we propose a novel online learning algorithm OLVF to learn from data with arbitrarily varying feature spaces. The OLVF algorithm learns to classify the feature spaces and the instances from feature spaces simultaneously. To classify an instance, the algorithm dynamically projects the instance classifier and the training instance onto their shared feature subspace. The feature space classifier predicts the projection confidences for a given feature space. The instance classifier will be updated by following the empirical risk minimization principle and the strength of the constraints will be scaled by the projection confidences. Afterwards, a feature sparsity method is applied to reduce the model complexity. Experiments on 10 datasets with varying feature spaces have been conducted to demonstrate the performance of the proposed OLVF algorithm. Moreover, experiments with trapezoidal data streams on the same datasets have been conducted to show that OLVF performs better than the state-of-the-art learning algorithm (Zhang et al. 2016).

Online modeling and identification of plug-in electric vehicles sharing a residential station

In this paper, a novel plug-in electric vehicle (PEV) modeling approach is proposed for residential charging stations. This methodology can be used in the design of autonomous energy management systems (EMS) with the purpose of providing the charging schedules using minimum input measured data. The proposed method is composed of two main steps. In the first step, unlike other similar works, online PEV recognition is performed by means of an artificial neural network as a supervised classification method. The required feature space for classification is provided using the power spectral density estimation and the statistical analysis of individual PEVs charging current. The second step deals with the statistical modeling of the charging habits to facilitate the scheduling by predicting the charging demand using the current measurements on the grid side. In this respect, each PEV charging habit is modeled based on the correlation among the plug-in time, departure time, required energy, and weekdays using kernel density estimation. The performance of the suggested method is validated using real data collected from a charging station. The final results confirm the applicability of the proposed methodology with a satisfactory precision. The effectiveness of the method is demonstrated using a comparative analysis in terms of the recognition performance.