China
Africa
Explore chapters and articles related to this topic
A Comprehensive Study on MLP and CNN, and the Implementation of Multi-Class Image Classification using Deep CNN
Published in K. Gayathri Devi, Kishore Balasubramanian, Le Anh Ngoc, Machine Learning and Deep Learning Techniques for Medical Science, 2022
CNN is extensively used for image classification and object detection. It has several advantages over MLP. They are,CNN's are capable of working with image data in 2D. So, there is no need to flatten the input images to 1D like MLP. It helps in retaining the spatial properties of images.It finds the relevant traits without the need for human interventionIt converges faster than the MLP modelIt has the highest accuracy than MLP in image predictionIt is also computationally efficient
Deep Learning for Medical Dataset Classification Based on Convolutional Neural Networks
Published in R. Sujatha, S. L. Aarthy, R. Vettriselvan, Integrating Deep Learning Algorithms to Overcome Challenges in Big Data Analytics, 2021
Figure 8.2 illustrates the classification framework based on CNNs. In order to achieve the classification of the dataset, the dataset first needs to be preprocessed; then, the data is selected; the data is cleaned; and, finally, the feature extraction method is used (Yadav & Jadhav, 2019). A CNN-based method has been used broadly in medical classification systems. When processing on small datasets, a CNN has the capability of classification with a higher performance. To improve the performance of using CNN methods, there are a few different strategies, including data augmentation and transfer learning (Sarraf & Tofighi, 2016). SVM classifiers are also used in the transfer learning of VGG-16 and Inception-V3 because SVMs also provide efficient performance. Using data augmentation prevents the overfitting of the network model.
Biometric Monitoring in Healthcare IoT Systems Using Deep Learning
Published in Sudhir Kumar Sharma, Bharat Bhushan, Narayan C. Debnath, IoT Security Paradigms and Applications, 2020
Shefali Arora, Veenu, M.P.S. Bhatia, Gurjot Kaur
Convolutional neural networks (convnets, CNNs) are mainly used for image classification. Originally developed by Geoffery Hinton, they are ideal in detecting objects in different positions in different images. These networks make use of convolution. Given 1D input x and a filter k, this operation is defined as: y[n]=(x*k)[n]∑−∞∞x[m]k[n−m]
Deep learning framework for biometric authentication using retinal images
Published in Computer Methods in Biomechanics and Biomedical Engineering: Imaging & Visualization, 2023
Jarina B. Mazumdar, S. R. Nirmala
The objective of the proposed work is to establish a robust retinal identification using colour retinal images as input. We have adopted a CNN model to solve this problem. CNN is a powerful architecture which is extensively used in image processing and pattern recognition. The structure of CNN includes a feature extraction layer where the input of each neuron is connected to the local receptive field of the former layer and extracts the local feature. The next layer is feature map layer, where features are calculated as convolution between receptive field and the filter. The outputs are collected as feature map. Furthermore, the filters in the same feature map plane have identical weight, so that the entire input image can be studied with the same feature detector. This is the benefit of convolution network with respect to the neuronal network connected to each other. Furthermore, the structure of feature map uses the sigmoid function as an activation function of the convolution network, which makes the feature map shift invariant. This makes CNN advantageous over classical machine learning approaches.
A Deep Learning Approach for Classification of Dentinal Tubule Occlusions
Published in Applied Artificial Intelligence, 2022
Anday Duru, İsmail Rakıp Karaş, Fatih Karayürek, Aydın Gülses
The architecture of CNN was biologically inspired by the organization of the Visual Cortex (Fukushima 1988; Hubel and Wiesel 1968). Particularly, the structure of the visual system behaves in such a way as to encode visual relations layer by layer. Each layer gradually represents more specific features (Hubel and Wiesel 1962). CNN is designed to emulate the same organizational behavior. CNN is a type of deep learning model, and they are one of the most powerful tools to do image recognition, classifications, segmentation, and many more tasks (Girshick et al. 2014; Ren et al. 2017; Sermanet et al. 2014). Each input image passes through a series of building blocks to classify relevant objects. Building blocks are the sequence of layers used to assign weights and biases to neurons.
Prostate cancer classification with MRI using Taylor-Bird Squirrel Optimization based Deep Recurrent Neural Network
Published in The Imaging Science Journal, 2022
Goddumarri Vijay Kumar, Mohammed Ismail Bellary, Thota Bhaskara Reddy
The pre-processed result is the noise-free image , which is passed as input to the segmentation stage for partitioning the image into a number of segments using the CNN approach [47,48]. CNN is comprised with the number of self-optimized neurons. It is the deep learning network model that automatically divides the image into different segments. For tasks including object detection, image segmentation, face recognition, and others, CNN is a frequently utilized technique. The CNN effectively completes the segmentation process since it is made up of multilayers connected to neurons. CNN learns the pre-processed output and segments the image automatically as a result of its transfer learning and automated features. The CNN structure consists of the input layer, hidden layer, and output layer. The hidden layer of this network topology consists of the convolutional layer, the pooling layer, and the fully connected layer.