Learning Method For Image Classification Research Articles

VisActive: Visual-concept-based Active Learning for Image Classification under Class Imbalance

Active learning methods recommend the most informative images from a large unlabeled dataset for manual labeling. These methods improve the performance of an image classifier while minimizing manual labeling efforts. We propose VisActive, a visual-concept-based active learning method for image classification under class imbalance. VisActive learns a visual concept, a generalized representation that holds the most important image characteristics for class prediction, and then recommends for each class four sets of unlabeled images with different visual concepts to increase the diversity and enlarge the training dataset. Experimental results on four datasets show that VisActive outperforms the state-of-the-art deep active learning methods.

FedUSC: Collaborative Unsupervised Representation Learning From Decentralized Data for Internet of Things

Federated Learning (FL) lately has shown much promise in improving the shared model and preserving data privacy. However, these existing methods are only of limited utility in the Internet of Things (IoT) scenarios, as they either heavily depend on high-quality labeled data or only perform well under idealized conditions, which typically cannot be found in practical applications. In this paper, we propose a novel federated unsupervised learning method for image classification without the use of any ground truth annotations. In IoT scenarios, a big challenge is that decentralized data among multiple clients is normally non-IID, leading to performance degradation. To address this issue, we further propose a dynamic update mechanism that can decide how to update the local model based on weights divergence. Extensive experiments show that our method outperforms all baseline methods by large margins, including +6.67% on CIFAR-10, +5.15% on STL-10, and +8.44% on SVHN in terms of classification accuracy. In particular, we obtain promising results on Mini-ImageNet and COVID-19 datasets and outperform several federated unsupervised learning methods under non-IID settings.

Class-Incremental Learning: Survey and Performance Evaluation on Image Classification.

For future learning systems, incremental learning is desirable because it allows for: efficient resource usage by eliminating the need to retrain from scratch at the arrival of new data; reduced memory usage by preventing or limiting the amount of data required to be stored - also important when privacy limitations are imposed; and learning that more closely resembles human learning. The main challenge for incremental learning is catastrophic forgetting, which refers to the precipitous drop in performance on previously learned tasks after learning a new one. Incremental learning of deep neural networks has seen explosive growth in recent years. Initial work focused on task-incremental learning, where a task-ID is provided at inference time. Recently, we have seen a shift towards class-incremental learning where the learner must discriminate at inference time between all classes seen in previous tasks without recourse to a task-ID. In this paper, we provide a complete survey of existing class-incremental learning methods for image classification, and in particular, we perform an extensive experimental evaluation on thirteen class-incremental methods. We consider several new experimental scenarios, including a comparison of class-incremental methods on multiple large-scale image classification datasets, an investigation into small and large domain shifts, and a comparison of various network architectures.

Variable-length deep convolutional neural networks by Internet Protocol Addresses Whale Optimization Algorithm for random and complex image classification

While Deep Convolutional Neural Network (DCNN) is one of the most reliable deep learning methods for image classification, choosing the optimal DCNN architecture for a particular application can be quite challenging. This study aims to examine the usage of the Whale Optimization Algorithm (WOA) to discover the optimal structure for DCNNs automatically. Three advancements based on standard WOA are proposed to accomplish the objective. First, a novel Internet Protocol Address (IPA)-based coding approach is suggested, which makes it easier to encode DCNN layers utilizing whale vectors. The construction of variable-length DCNNs is then suggested using an Enfeebled layer to include certain whale vector dimensions. The learning procedure concludes by subdividing large datasets into smaller portions which are then arbitrarily assessed. In order to evaluate the proposed model’s effectiveness, it is compared to the performance of twenty-three different classifiers, including the state-of-the-art algorithm, on nine standard image classification benchmark datasets. The experimental results indicate that the suggested technique outperforms other benchmarks in 81 of 95 assessments. This variable-length approach is the first of its kind, utilizing WOA to evolve the topologies of DCNNs automatically.

Multi‐view convolutional neural network‐based target classification in high‐resolution automotive radar sensor

In this study, a target classification method based on point cloud data in a high-resolution radar sensor is proposed. By using multiple antenna elements arranged in horizontal and vertical directions, pedestrians, cyclists and vehicles can be expressed as point cloud data in the three-dimensional (3D) space. To perform target classification using the spatial characteristics (i.e. length, height and width) of the target, the 3D point cloud data is orthogonally projected onto the xy, yz and zx planes, respectively, and three types of images are generated. Then, a multi-view convolutional neural network (CNN)-based target classifier using those three images as inputs is designed. To this end, a method for synthesising the detection results of three directions in series or in parallel is proposed. The proposed classifier can learn the spatial characteristics of the target by using the detection results of multiple viewpoints. Compared to the CNN-based classifier that uses only the detection result of a single plane as input, the proposed method shows 4.5%p higher classification accuracy in terms of the target with the lowest classification accuracy. In addition, the proposed multi-view CNN structure shows improved classification performance and shorter training time compared to the well-known deep learning methods for image classification.

Klasifikasi Citra Daging Sapi dan Daging Babi Menggunakan Ekstraksi Ciri dan Convolutional Neural Network

Cases of mixing beef and pork are still happening today. The increasing demand for beef causes many traders to mix meat to gain more profit. Distinguishing beef and pork can be done by sight and smell, but still has weaknesses. This study uses Deep Learning method for image classification with Convolutional Neural Network architecture EfficientNet-B0. The amount of data is 3,000 images which are divided into 3 classes, beef, pork, and mixed meat. This study uses original image data and image data of Contrast Limited Adaptive Histogram Equalization. The data is divided by the ratio of training data and test data of 80:20. The results of testing the model with the confusion matrix show the highest classification performance with 95.17% accuracy, 92.72% precision, 95.5% recall, and 94.09% f1 score, in the original image data with the use of neurons in the first dense amounting to 256, 32 batch size, 0.01 learning rate, and Adam's optimizer

A novel discriminative dictionary learning method for image classification

A novel discriminative dictionary learning method for image classification

Convex hull-based distance metric learning for image classification

Distance metric learning is an important topic in visual classification tasks. Learning an appropriate distance measure can greatly improve the performance of image recognitions. In this paper, we model an image set as a regularized convex hull and propose a distance metric learning method for image classification based on regularized convex hulls. In particular, a regularized point-to-convex hull distance metric (RPCHD) and a regularized convex hull-to-convex hull distance metric (RCHCHD) are introduced to measure the distance between the query image object and the existing image sets for image classification tasks. The coefficients in distance metric are solved by an approximate optimization strategy. Two SVM-like distance metric learning models are constructed and transformed into the standard support vector machines to learn distance metric matrix in RPCHD and RCHCHD. Positive and negative sample pairs are proposed to represent distance metric matrix. Experiments on three image databases show that our proposed RPCHD and RCHCHD can effectively improve the performance of image classification.

Activation ensemble generative adversarial network transfer learning for image classification

Transfer learning provides a useful solution to learn a new conceptual domain from few examples, which exploits prior knowledge from a related domain. We proposed a simple and yet effective transfer learning method for image classification that constructs an activation ensemble generative adversarial net (AE-GAN) to transfer knowledge from one dataset to another. The AE-GAN is mainly composed of three convolutional layers and adopts an ensemble of multiple activation functions. Experimental results on five benchmark datasets show that when only a few samples are available for training a target task, leveraging datasets from other related datasets by AE-GAN can significantly improve the performance for image classification with a small set of samples.

Manifold Transfer Learning via Discriminant Regression Analysis

In transfer learning, how to effectively transfer useful information from the source domain to the target domain is crucial. In this paper, we propose a novel transfer learning method for image classification, named manifold transfer learning via discriminant regression analysis (MTL-DRA), to transfer the local geometry structure information from the source domain to the target domain and ensure that the transform matrix is robust or sparse so that samples from different domains can be well combined. In MTL-DRA, we encode discriminant information of the source domain to the target domain by introducing between- and within-class graphs to preserve within-class similarity and reduce between-class similarity. With different norms as constraints, MTL-DRA overcomes the disturbance of noise and avoids negative transfer learning. To improve the robustness of MTL-DRA, we encode a nuclear norm constraint and propose robust MTL-DRA (RMTL-DRA). We analyzed the convergence and complexity of the two proposed methods. To verify the performance of the proposed methods, we conducted extensive experiments on five public image benchmarks. The experimental results show that the proposed methods outperform state-of-the-art transfer learning methods.

Exploring the role of deep neural networks for post-disaster decision support

Disaster management operations are information intensive activities due to high uncertainty and complex information needs. Emergency response planners need to effectively plan response activities with limited resources and assign rescue teams to specific disaster sites with high probability of survivors swiftly. Decision making becomes tougher since the limited information available is heterogenous, untimely and often fragmented. We address the problem of lack of insightful information of the disaster sites by utilizing image data obtained from smart infrastructures. We collect geo-tagged images from earthquake-hit regions and apply deep learning method for classification of these images to identify survivors in debris. We find that deep learning method is able to classify the images with significantly higher accuracy than the conventionally used machine learning methods for image classification and utilizes significantly lesser time and computational resources. The novel application of image analytics and the resultant findings from our models have valuable implications for effective disaster response operations, especially in smart urban settlements.

A Semi-Supervised CNN With Fuzzy Rough C-Mean for Image Classification

While deep learning (DL) has proven to be a powerful paradigm for the classification of the large-scale big image data set. Deep learning network such as CNN requires a large number of labeled samples for training the network. However, often times the labeled data are difficult, expensive, and time-consuming. In this paper, we propose a semi-supervised approach to fused fuzzy-rough C-mean clustering with convolutional neural networks (CNNs) to knowledge learning from simultaneously intra-model and inter-model relationships forming final data representation to be classified, which can be achieved by better performance. Conception behind this is to reduce the uncertainty in terms of vagueness and indiscernibility by using fuzzy-rough C-mean clustering and specifically removing noise samples by using CNN from raw data. The framework of our proposed semi-supervised approach uses all the training data with abundant unlabeled data with few labeled data to train FRCNN model. To show the effectiveness of our model, we used four benchmark large-scale image datasets and also compared it with the state-of-the-art supervised, unsupervised, and semi-supervised learning methods for image classification.

Abnormality Detection in Musculoskeletal Radiographs Using Capsule Network

To treat the diseases or injuries of the joints, bones, muscles, and spine in both adult and pediatric imaging the musculoskeletal radiographs bring a significant depth of expertise. Abnormality detection in the musculoskeletal study is backbreaking as more than 1.7 billion people are affected by musculoskeletal condition (BMU, 2017). Hence if we want to create enough opportunity to treat a maximum amount of patients, machine learning and deep learning can play a crucial role. CNN is an excellent deep learning method for image classification and other computer vision tasks. But CNN has exhibited some serious limitations when the images are rotated and deformed. Hence capsule network architecture is introduced in this paper for musculoskeletal radiographs abnormality detection and this capsnet architecture has shown very promising features that can help to vanquish the limitations of CNN. In addition, this capsule network has scored 10% higher kappa score than 169 layer densenet using less training data in the case of musculoskeletal radiographs abnormality detection. This feature of capsule network can help to use deep learning in such cases where an aggregate of a large amount of data is not possible. For image quality investigation, blind image spatial quality evaluator (BRISQUE) and naturalness image quality evaluator (NIQE) scores are measured and it is found that when the pixel size of the resized images are more close to the pixel size of the original images, we get a better approximation. Hence in the case of musculoskeletal radiographs abnormality detection, our method outperforms state-of-the-art method using a less amount of training data.

Multi-pixels Classification for nuclei segmentation in digital pathology based on deep machine learning

The problem this paper solves is identifying the accurate boundary for nuclei in breast cancer digital pathology images. By doing the nuclei segmentation, the output image can be a reference for the doctors to make better diagnosis for breast cancer patients. Using convolutional neural network to solve this problem which is a popular deep learning method for image classification. By generating labelled patches from the original dataset for training and using a sliding window to cut out patches from testing image for classifying pixels in the testing image, the image level classification can be transferred to pixel level classification. A 2x2 receptive field can be classified to 16 classes, and each pixel can be calculated 4 time, which increase the pixel level TPR (true positive rate) from 0.868 to 0.909. And the boundaries are smother than traditional method that the receptive field is 1x1.

Neural-Response-Based Extreme Learning Machine for Image Classification.

This paper proposes a novel and simple multilayer feature learning method for image classification by employing the extreme learning machine (ELM). The proposed algorithm is composed of two stages: the multilayer ELM (ML-ELM) feature mapping stage and the ELM learning stage. The ML-ELM feature mapping stage is recursively built by alternating between feature map construction and maximum pooling operation. In particular, the input weights for constructing feature maps are randomly generated and hence need not be trained or tuned, which makes the algorithm highly efficient. Moreover, the maximum pooling operation enables the algorithm to be invariant to certain transformations. During the ELM learning stage, elastic-net regularization is proposed to learn the output weight. Elastic-net regularization helps to learn more compact and meaningful output weight. In addition, we preprocess the input data with the dense scale-invariant feature transform operation to improve both the robustness and invariance of the algorithm. To evaluate the effectiveness of the proposed method, several experiments are conducted on three challenging databases. Compared with the conventional deep learning methods and other related ones, the proposed method achieves the best classification results with high computational efficiency.

Patch-level Tumor Classification in Digital Histopathology Images with Domain Adapted Deep Learning.

Tumor histopathology is a crucial step in cancer diagnosis which involves visual inspection of imaging data to detect the presence of tumor cells among healthy tissues. This manual process can be time-consuming, error-prone, and influenced by the expertise of the pathologist. Recent deep learning methods for image classification and detection using convolutional neural networks (CNNs) have demonstrated marked improvements in the accuracy of a variety of medical imaging analysis tasks. However, most well-established deep learning methods require large annotated training datasets that are specific to the particular problem domain; such datasets are difficult to acquire for histopathology data where visual characteristics differ between different tissue types, in addition to the need for precise annotations. In this study, we overcome the lack of annotated training dataset in histopathology images of a particular domain by adapting annotated histopathology images from different domains (tissue types). The data from other tissue types are used to pre-train CNNs into a shared histopathology domain (e.g., stains, cellular structures) such that it can be further tuned/optimized for a specific tissue type. We evaluated our classification method on publically available datasets of histopathology images; the accuracy and area under the receiver operating characteristic curve (AUC) of our method was higher than CNNs trained from scratch on limited data (accuracy: 84.3% vs. 78.3%; AUC: 0.918 vs. 0.867), suggesting that domain adaptation can be a valuable approach to histopathological images classification.

Image Classification With Tailored Fine-Grained Dictionaries

In this paper, we propose a novel fine-grained dictionary learning method for image classification. To learn a high-quality discriminative dictionary, three types of multispecific subdictionaries, i.e., class-specific dictionaries (CSDs), universal dictionary (UD), and family-specific dictionaries (FSDs), are simultaneously uncovered. Here, CSDs and UD, respectively, model the patterns for each class and the patterns irrespective of any class. FSDs can help reveal the shared patterns between multiple image classes, by filling the gap between the patterns in CSDs and UD. The dependence among image classes is revealed by the shared FSDs, and a common FSD can be assigned to several classes to represent their residual. Finally, the most discriminative FSD for each class is identified by minimizing the sparse reconstruction error. Extensive experiments are conducted on different widely used data sets for image classification. The results demonstrate the superior performance of the proposed method over some state-of-the-art methods.

Balanced Active Learning Method for Image Classification

The manual labeling of natural images is and has always been painstaking and slow process, especially when large data sets are involved. Nowadays, many studies focus on solving this problem, and most of them use active learning, which offers a solution for reducing the number of images that need to be labeled. Active learning procedures usually select a subset of the whole data by iteratively querying the unlabeled instances based on their predicted informativeness. One way of estimating the information content of an image is by using uncertainty sampling as a query strategy. This basic technique can significantly reduce the number of label needed; e.g. to set up a good model for classification. Our goal was to improve this method by balancing the distribution of the already labeled images. This modification is based on a novel metric that we present in this paper. We conducted experiments on two popular data sets to demonstrate the efficiency of our proposed balanced active learning (BAL) approach, and the results showed that it outperforms the basic uncertainty sampling.

A survey of deep learning methods and software tools for image classification and object detection

Deep learning methods for image classification and object detection are overviewed. In particular we consider such deep models as autoencoders, restricted Boltzmann machines and convolutional neural networks. Existing software packages for deep learning problems are compared.

Semi-Supervised Image Classification Based on Local and Global Regression

The insufficiency of labeled samples is a major problem in automatic image annotation. However, unlabeled samples are readily available and abundant. Hence, semi-supervised learning methods, which utilize partly labeled samples and a large amount of unlabeled samples, have attracted increased attention in the field of image classification. During the past decade, graph-based semi-supervised learning became one of the most important research areas in semi-supervised learning. In this letter, we propose a novel and effective graph based semi-supervised learning method for image classification. The new method is based on local and global regression regularization. The local regression regularization adopts a set of local classification functions to preserve both local discriminative and geometrical information; while the global regression regularization preserves the global discriminative information and calculates the projection matrix for out-of-sample extrapolation. Extensive simulations based on synthetic and real-world datasets verify the effectiveness of the proposed method.