Key Publications

Submitted and Published Papers

Triage of potential COVID-19 patients from chest X-ray images using hierarchical convolutional networks

The current COVID-19 pandemic has motivated the researchers to use artificial intelligence techniques for a potential alternative to reverse transcription-polymerase chain reaction due to the limited scale of testing. The chest X-ray (CXR) is one of the alternatives to achieve fast diagnosis, but the unavailability of large-scale annotated data makes the clinical implementation of machine learning-based COVID detection difficult. Another issue is the usage of ImageNet pre-trained networks which does not extract reliable feature representations from medical images. In this paper, we propose the use of hierarchical convolutional network (HCN) architecture to naturally augment the data along with diversified features. The HCN uses the first convolution layer from COVIDNet followed by the convolutional layers from well-known pre-trained networks to extract the features. The use of the convolution layer from COVIDNet ensures the extraction of representations relevant to the CXR modality.

Genetic algorithm based weighted extreme learning machine for binary imbalance learning

Abstract:

Class imbalance problem refers to unequal distribution of data instances between classes. Due to this, popular classifiers misclassify data instances of minority class into majority class. Initially, Extreme learning machine was proposed with the prime objective of handling real valued datasets. Though, it a fast learning technique, it suffers from the drawback of misclassification of imbalanced dataset which leads to the class imbalance problem. So, a new variant of ELM called Weighted Extreme Learning Machine was developed. This technique aimed at handling imbalance data by assigning more weight to minority class and less weight to majority class. The limitation of this technique lied in that it generates weight according to class distribution of training data, thereby, creating dependency on input data. This leads to the lack of finding optimal weight at which good generalization performance could be achieved. This work uses Genetic Algorithm to find optimal weight which is given to minority and majority class instances.

Published in: 2015 International Conference on Cognitive Computing and Information Processing(CCIP)

Date of Conference: 3-4 March 2015

Date Added to IEEE Xplore: 04 May 2015

Electronic ISBN:978-1-4799-7171-8

INSPEC Accession Number: 15109728

DOI: 10.1109/CCIP.2015.7100711

Publisher: IEEE Conference Location: Noida, India

Prediction of Ubiquitination Sites Using UbiNets

Abstract

Ubiquitination controls the activity of various proteins and belongs to posttranslational modification. Various machine learning techniques are taken for prediction of ubiquitination sites in protein sequences. The paper proposes a new MLP architecture, named UbiNets, which is based on Densely Connected Convolutional Neural Networks (DenseNet). Computational machine learning techniques, such as Random Forest Classifier, Gradient Boosting Machines, and Multilayer Perceptrons (MLP), are taken for analysis. The main target of this paper is to explore the significance of deep learning techniques for the prediction of ubiquitination sites in protein sequences. Furthermore, the results obtained show that the newly proposed model provides significant accuracy. Satisfactory experimental results show the efficiency of proposed method for the prediction of ubiquitination sites in protein sequences. Further, it has been recommended that this method can be used to sort out real time problems in concerned domain.

Publications: Publications