Image Classification using Deep Learning with Support Vector Machines (original) (raw)
Related papers
INFOCOMP Journal of Computer Science, 2019
Classification is the technique applied in data mining to form groups under specified class labels. Classification is supervised type of machine learning. In this paper, two popular classification techniques, Support Vector Machine (SVM) and Convolutional Neural Network (CNN) are compared for accuracy of classification of images. Image classification is done on the basis of feature selection and feature extraction using Tensor flow package. The two classifiers understudy are Linear (SVM) and non-linear techniques (CNN). CNN possesses a powerful feature extraction and SVM is considered as a high-end classifier. Complexity of the feature extraction and selection can be increased in CNN by adding more layers, but in SVM complexity cannot be increased. CNN processes images using matrices of weights and is called as filters or features that detect specific attributes such as vertical edges, horizontal edges, etc. As and when the image progresses through each layer, the filters can recognize more and more complex attributes. In this proposed study, graphs of training phase of CNN also show how the training results are improved image by image due to increasing knowledge of features and thus loss is decreased. This research study focuses on accuracy measure of the above mentioned methods. For the image classification studied in this paper, it has been observed that SVM gives adequate accuracy for binary classification whereas CNN gives consistent accuracy over binary as well as multi class classification problems. The recognition rate achieved by the CNN algorithm varies between 75%-75.40 % for binary and multiclass classification. SVM accuracy rate decreases from 80.95 % for binary classification, to 50 % for multiclass classification.
Support vector machines (SVMs) have been around for decades, they have been used for a number of classification tasks. They actually have a very strong theory behind them, which make it relatively easy to choose the best hyper-parameters. The kernel trick makes it easier for SVMs to implicitly classify in higher dimensional space, making it possible to work with nonlinearly separable datasets. On the other hand, Convolutional Neural Networks (CNNs) have gained important attention in recent years for their high performance in image classification problems with high number of categories. The automatic feature extraction of convolutional layer and the dimensionality reduction of the pooling layer make CNN gain high predictive power on testing data. In this work both models are briefly discussed and implemented on a binary classification problem from the EMIST character dataset. The CNN outperformed SVM achieving a misclassification error rate on test data of 1.7 % against 2.32 % for SVM.
Deep Learning using Linear Support Vector Machines
Recently, fully-connected and convolutional neural networks have been trained to achieve state-of-the-art performance on a wide variety of tasks such as speech recognition, image classification, natural language processing , and bioinformatics. For classification tasks, most of these " deep learning " models employ the softmax activation function for prediction and minimize cross-entropy loss. In this paper, we demonstrate a small but consistent advantage of replacing the soft-max layer with a linear support vector machine. Learning minimizes a margin-based loss instead of the cross-entropy loss. While there have been various combinations of neu-ral nets and SVMs in prior art, our results using L2-SVMs show that by simply replacing softmax with linear SVMs gives significant gains on popular deep learning datasets MNIST, CIFAR-10, and the ICML 2013 Representation Learning Workshop's face expression recognition challenge.
Deep Features for Training Support Vector Machines
Journal of Imaging, 2021
Features play a crucial role in computer vision. Initially designed to detect salient elements by means of handcrafted algorithms, features now are often learned using different layers in convolutional neural networks (CNNs). This paper develops a generic computer vision system based on features extracted from trained CNNs. Multiple learned features are combined into a single structure to work on different image classification tasks. The proposed system was derived by testing several approaches for extracting features from the inner layers of CNNs and using them as inputs to support vector machines that are then combined by sum rule. Several dimensionality reduction techniques were tested for reducing the high dimensionality of the inner layers so that they can work with SVMs. The empirically derived generic vision system based on applying a discrete cosine transform (DCT) separately to each channel is shown to significantly boost the performance of standard CNNs across a large and ...
Deep support vector neural networks
Integrated Computer-Aided Engineering, 2020
Kernel based Support Vector Machines, SVM, one of the most popular machine learning models, usually achieve top performances in two-class classification and regression problems. However, their training cost is at least quadratic on sample size, making them thus unsuitable for large sample problems. However, Deep Neural Networks (DNNs), with a cost linear on sample size, are able to solve big data problems relatively easily. In this work we propose to combine the advanced representations that DNNs can achieve in their last hidden layers with the hinge and ϵ insensitive losses that are used in two-class SVM classification and regression. We can thus have much better scalability while achieving performances comparable to those of SVMs. Moreover, we will also show that the resulting Deep SVM models are competitive with standard DNNs in two-class classification problems but have an edge in regression ones.
Computer Vision: Classification of Images Based On Deep Learning with the CNN Architecture Model
International Journal of Engineering Research in Computer Science and Engineering
Deep learning is a scientific field in Machine Learning (ML) that is developing with various applications, one of which is visual image processing technology. With the excellent capabilities of computer vision, image processing from computer visuals is used to duplicate the human ability to understand object information in the image. One of the Machine Learning (ML) methods that can be used for object classification in images is the Convolution Neural Network (CNN) method. The two core stages when processing object classification in the image, the first stage is image classification using feedforward, and the second stage applies the backpropagation method. In this study, before the classification stage, this method was first carried out through preprocessing, which is useful as an image separation to focus on the object to be classified. Furthermore, it is carried out by conducting pre-training using the feedforward method with the bias weights, which are updated after every traini...
Image Classification using Deep Learning
IRJET, 2022
Deep Learning aims to work on complex data and achieve accuracy. It works on AI-based domains like Natural Language Processing and Computer vision[1]. In deep learning, the computer model learns to perform classification of task from images, text or sound. Image classification is the task of extracting essential features from given input image that are required to predict the correct classification. The objective is to build a Convolution Neural Network model that can correctly predict and classify the input image as Dog or Cat. The classification is done by extracting specific features of the input image. The CNN Model consists of various layers like Convolution layer, ReLU layer, Pooling layer, etc. The model is trained well with training data. At last, the CNN model is tested for accuracy in image classification with the help of some test images.
Classification of Images Using CNN Model and its Variants
IRJET, 2023
Image classification is a method of assigning a label to an image and it is suitable to use deep learning for this task due to spatial nature of image which can leverage the massively parallel structure to learn various features. In this research, a Convolution Neural Networks (CNN) model is presented with three configurations. The first configuration is simple and other two configurations are improvement of first configuration by using techniques to prevent over fitting. The training and testing is performed on a CIFAR-10 dataset which consists of 60000 sets of images of 10 different objects. During comparison of variants of model using different performance matrices it is observed that the dropout regularization technique can significantly make the model more accurate. It also shows that lower batch size can give better result than higher batch size.
Convolutional Neural Networks and Pattern Recognition: Application to Image Classification
2019
This research study focuses on pattern recognition using<br> convolutional neural network. Deep neural network has been<br> choosing as the best option for the training process because<br> it produced a high percentage of accuracy. We designed<br> different architectures of convolutional neural network in<br> order to find the one with high accuracy of image<br> classification and optimum bias. We used CIFAR-10 data<br> set that contains 60000 Images to train our model on<br> architectures. The best architecture was able to classify<br> images with 95.55% of accuracy and an error of 0.32%<br> using cross validation method. We note that, the numbers of<br> epoch while running the model and the depth of the<br> architecture are factors that contributed to get this<br> performance.
Image Classification Using Deep Learning Network
Deep learning methods are revolutionizing the image classification. CNN are the most successful method in deep learning because conventional image classification technique were based on the hand coded features which were not robust to different lightning conditions and would fail when exposed to different object orientation. In this paper an architecture is selected after comparing different architectures on the dataset which is also optimized to predict the object belonging to the classes in the dataset. Finally the analysis is given along with conclusions