The IAES’s Nawala: Deep Learning Technology

Greetings, fellow Nawala! May you always be in good health.

This is the IAES Nawala of the  Institute of Advanced Engineering and Science. Today we would like to share an insight into deep learning (DL) technology. DL allows computers to recognize and extract visual characteristics. In the process of recognizing and extracting visual characteristics, computers use methods and techniques to achieve the desired accuracy of results. Convolutional neural network (CNN) is one of the methods used to process visual characteristics. Reddy and Khanaa, in their research, used CNN to detect and classify lung cancer based on image processing.  More details can be read in the following article:

Intelligent deep learning algorithm for lung cancer detection and classification

N. Sudhir Reddy, V. Khanaa

Lung cancer is one of the leading causes of cancer mortality. The overlapping of cancer cells makes early diagnosis difficult. When lung cancer is found early, many therapy choices are reduced, the danger of invasive surgery is reduced, and the chance of survival increases. The primary goal of this study work is to identify early-stage lung cancer and categories using an intelligent deep learning algorithm. Following a thorough review of the literature, we discovered that certain classifiers are ineffective while others are almost perfect. In general, several different kinds of images are employed, but computer tomography scanned images are preferable due to their reduced noise. Intelligent deep learning algorithm is one such approach that employs convolutional neural network techniques and has been shown to be the most effective way for medical image processing, lung nodule identification, classification, feature extraction, and lung cancer prediction. The characteristics are taken from the segmented images and classified using intelligent deep learning algorithm. The suggested techniques’ performances are assessed based on their accuracy, sensitivity, specificity, recall, and precision.

CNN continues to be developed to produce higher accuracy. Jasim and Atia developed a block-based CNN to classify images. The results obtained from the proposed method can improve accuracy compared to other CNN methods by 3%. More detailed information can be read in the following article:

Towards classification of images by using block-based CNN

Retaj Matroud Jasim, Tayseer Salman Atia

Image classification is the process of assigning labeling to the input images to a fixed set of categories; however, assigning labels to the image is difficult by using the traditional method because of the large number of images. To solve this problem, we will resort to deep learning techniques. Which is enables computers to recognize and extract visual characteristics. The convolutional neural network (CNN) is a deep neural network used for many purposes, such as image classification, detection, and face recognition, due to its high-performance accuracy in classification and detection tasks. In this paper, we develop CNN based on the transfer learning approach for image classification. The network comprises two types of transfer learning, ResNet and DenseNet, as building blocks of the network with an multilayer perceptron (MLP) classifier. The proposed method does not need to preprocess before these datasets that input into the network. It was train on two datasets: the Cifar-10 and the Sign-Traffic datasets. We conclude that the proposed method achieves the best performance compared with other states of the art. The accuracy gained is 97.45% and 99.45%, respectively, where the proposed CNN increased the accuracy compared to other methods by 3%.

In the medical world, DL development is essential. As in previous research that discussed the classification of lung cancer images, Kadhim and Kamil conducted a study that discussed breast cancer. The study combined DL with the Gabor filtering method to improve the accuracy of breast cancer diagnosis results. A detailed explanation can be read in the following article:

Breast invasive ductal carcinoma diagnosis using machine learning models and Gabor filter method of histology images

Rania R. Kadhim, Mohammed Y. Kamil

Breast cancer is the most common type of cancer in women and the leading cause of death from a malignant growth in the world. Machine learning methods have been created to help with cancer detection accuracy. There are several methods for detecting cancer. Histopathological images are more accurate. In this study, we employed the Gabor filter to extract statistical features from invasive ductal carcinoma histopathology images. From the histopathological images, we chose 100, 200, 400, 1000, and 2000 at random. These statistical features were used to train several models to classify these images as malignant or benign, including the decision tree, quadratic discriminant analysis, extra randomized trees, gradient boosting, Gaussian process, Naive Bayes, nearest centroid, multilayer perceptron, and support vector machine. The models’ accuracy, sensitivity, specificity, precision, and F1_score were examined. The models produced the highest results when there were 100 images and a wavenumber of 0.2. While as the number of images increased, the models’ effectiveness reduced. The most obvious finding to emerge from this study is that we suggest using deep learning instead of machine learning models for large datasets.

In addition to medicine, DL has also penetrated the psychology world. In research conducted by Agarwal et al., DL was implemented to classify human emotions. The data classified is not image data but music data that is listened to. The classification results can identify the emotions the music listener feels, such as dramatic, happy, aggressive, sad, and romantic. More details about the research can be read in full through the following link:

Emotion classification for musical data using deep learning techniques

Gaurav Agarwal, Sachi Gupta, Shivani Agarwal, Atul Kumar Rai

This research is done based on the identification and thorough analyzing musical data that is extracted by the various method. This extracted information can be utilized in the deep learning algorithm to identify the emotion, based on the hidden features of the dataset. Deep learning-based convolutional neural network (CNN) and long short-term memory-gated recurrent unit (LSTM-GRU) models were developed to predict the information from the musical information. The musical dataset is extracted using the fast Fourier transform (FFT) models. The three deep learning models were developed in this work the first model was based on the information of extracted information such as zero-crossing rate, and spectral roll-off. Another model was developed on the information of Mel frequencybased cepstral coefficient (MFCC) features, the deep and wide CNN algorithm with LSTM-GRU bidirectional model was developed. The third model was developed on the extracted information from Mel-spectrographs and untied these graphs based on two-dimensional (2D) data information to the 2D CNN model alongside LSTM models. Proposed model performance on the information from Mel-spectrographs is compared on the F1 score, precision, and classification report of the models. Which shows better accuracy with improved F1 and recall values as compared with existing approaches.

Some of the articles above are a small part of the research on deep learning development. To get more information please visit for FREE at: https://www.beei.org/ and https://ijres.iaescore.com/.

By: I. Busthomi