Akanksha Sharma

Bio: Akanksha Sharma is an academic researcher from Thapar University. The author has contributed to research in topics: Electric power system & Electricity market. The author has an hindex of 3, co-authored 14 publications receiving 43 citations.

Assessing alpha activity in attention and relaxed state: An EEG analysis

Abstract: An electroencephalogram is a recording of brains spontaneous electrical activity. This is controlled by billions of neurons. These neurons continually send messages to each other which can be picked up as electrical impulses from the scalp. The process of picking up and recording the impulses is known as EEG. An EEG can be divided into four basic frequency bands namely delta, theta, alpha and beta. This paper shows changes in the power of alpha frequency band on performing attention tasks. It is observed that mean alpha power decreased while performing attention task as compared to mean alpha power in relaxed state in the frontal and occipital region of the brain. No statistically significant changes in alpha power are found in prefrontal region of the brain during attention tasks.

Potential of generative adversarial net algorithms in image and video processing applications– a survey

Abstract: Generative Adversarial Network (GAN) has gained eminence in a very short period as it can learn deep data distributions with the help of a competitive process among two networks. GANs can synthesize images/videos from latent noise with a minimized adversarial cost function. The cost function plays a deciding factor in GAN training and thus, it is often subjected to new modifications to yield better performance. To date, numerous new GAN models have been proposed owing to changes in cost function according to applications. The main objective of this research paper is to present a gist of major GAN publications and developments in image and video field. Several publications were selected after carrying out a thorough literature survey. Beginning from trends in GAN research publications, basics, literature survey, databases for performance evaluation parameters are presented under one umbrella.

Comparison on Generative Adversarial Networks –A Study

Abstract: Various new deep learning models have been invented, among which generative adversarial networks have gained exceptional prominence in last four years due to its property of image synthesis. GANs have been utilized in diverse fields ranging from conventional areas like image processing, biomedical signal processing, remote sensing, video generation to even off beat areas like sound and music generation. In this paper, we provide an overview of GANs along with its comparison with other networks, as well as different versions of Generative Adversarial Networks.

Optimal Placement of TCSC for Congestion Management and Power Loss Reduction Using Multi-Objective Genetic Algorithm

Abstract: Electricity demand has been growing due to the increase in the world population and higher energy usage per capita as compared to the past. As a result, various methods have been proposed to increase the efficiency of power systems in terms of mitigating congestion and minimizing power losses. Power grids operating limitations result in congestion that specifies the final capacity of the system, which decreases the conventional power capabilities between coverage areas. Flexible AC Transmission Systems (FACTS) can help to decrease flows in heavily loaded lines and lead to lines loadability improvements and cost reduction. In this paper, total power loss reduction and line congestion improvement are assessed by determining the optimal locations and compensation rates of Thyristor-Controlled Series Compensator (TCSC) devices using the Multi-Objective Genetic Algorithm (MOGA). The results of applying the proposed method on the IEEE 30-bus test system confirmed the efficiency of the proposed procedure. In addition, to check the performance, applicability, and effectiveness of the proposed method, different heuristic algorithms, such as the multi-objective Particle Swarm Optimization (PSO) algorithm, Differential Evolution (DE) algorithm, and Mixed-Integer Non-Linear Program (MINLP) technique, are used for comparison. The obtained results show the accuracy and fast convergence of the proposed method over the other heuristic techniques.

EEG in classroom: EMD features to detect situational interest of students during learning

Abstract: Situational interest is widely explored in the psychology and education domains. It is proven to have positive effect on learning and academic achievement. Nonetheless, not much attention is given for assessing the feasibility of detecting this interest in natural classroom physiologically. Therefore, this study investigates the possibility of detecting situational interest using Electroencephalogram (EEG) in classroom. After preprocessing of EEG data, they were decomposed using Empirical Mode Decomposition (EMD). The resulted Intrinsic Mode Functions (IMFs) were ranked based on their significance using T-test and Receiver Operator Characteristics (ROC) in descending order. A matrix was constructed for all participants using the best six features from four EEG channels. These selected features were fed into Support Vector Machine (SVM) and K-Nearest Neighbor (KNN) classifiers with 10 cross validation. While SVM achieved high accuracy of 93.3% and 87.5% for two data sets using features from the four EEG channels, KNN classifier achieved high accuracy of 87.5% and 86.7% in the same datasets using single EEG channel. It is found that gamma and delta bands can be used successfully to detect situational interest of students during learning in classrooms. Furthermore, data of single EEG channel - F3 in this study- was efficient to detect student’s situational interest in simultaneous recording of EEG in classroom.