Mst Shapna Akter

I am a Research Assistant at North South University North South University, where I work on Deep learning field with real life application on medical image dataset, financial time series dataset, and text dataset under the supervision of Dr. Mahdy Rahman Chowdhury s. I also work on qualitative research field which includes Women in STEM and Women participation on technology and banking system(funded by Bill and Melindat gates foundation) under the supervision of Dr. Nova ahmed . I have also worked as a part-time research student at NSU Optics Lab for about two years(2019-2021).

Email  /  Linkedln /  google scholar

My interest lies in both technical and qualitative research work:
Technical field:

We have presented a large and diverse cell segmentation dataset that will promote future research towards clinically applicable cell segmentation methods from microscopic examinations, which can be later used for downstream tasks such as detecting hematological diseases (i.e., malaria). We have performed a comparative analysis on the dataset using several Machine learning and deep learning techniques.

The risk factor of frontier market has not been investigated yet. We have shown the process of predicting and forecasting the risk factor of this market. We have developed a novel stacking ensemble of the neural network model that performs best on multiple data patterns. We have compared our model’s performance with the performance results obtained by using some traditional machine learning ensemble models such as Random Forest, AdaBoost, Gradient Boosting Machine, and Stacking Ensemble, along with some traditional deep learning models such as Convolutional Neural Network (CNN), Long Short-Term Memory (LSTM) and Bidirectional Long Short-Term (BiLSTM) on twenty different companies dataset of a frontier market.

Data denoising or removing noise from a dataset is achallenging topic for the researchers. We have come up with a unique solution to filter out the noise of high magnitudes using several algorithms such as interpolation, extrapolation, spectral clustering, agglomerative clustering, wavelet analysis, and median filtering on the industrial dataset. We have also employed peak detection and peak validation algorithms to detect fuel refill and consumption in charge-discharge cycles.

Physical abuse and sexual harassment are seriousissues all over the world. In Bangladesh, India, and othersouth Asian countries, crimes of these natures have risen to asubstantial number during the past decade. We have designedan Arduino-based simple safety-vest device by using E-Textilesas pressure sensing fabric and incorporated a mobile applicationfor women. t has been designedwith the target of commercial use. The device has also beenbeta-tested to get some insights from the people who will be usingthis cost-effective device as a sense of protection

The main objective of this work is to forecast earthquake occurrences using time series models such as Holt-Winters. This work primarily focuses on the three distinct countries of the South Asian region, which are: Afghanistan, Nepal, and Bangladesh.

We did a comparative analysis on multiple skin cancer classes using deep convolutional neural networks, which will help the medical sector distinguish skin cancer from different skin lesions with high accuracy. Seven classes of skin lesions have been classified using Resnet-50, VGG-16, Densenet, Mobilenet, Inceptionv3, Xception, and CNN.

Autism is a dangerous disease that prevents children from communicating and interacting with others. We are trying to analyze the important features from autism child face image data that will help to detect the disease properly. The main idea is to develop appropriate filters for the important features and classify the ASD from non-ASD disease

This research aims to apply machine learning techniques to predict the volatility magnitude and analyze cryptocurrency risk factors. In twenty elements of cryptocurrency, machine learning algorithms and models have been used to make it easier for individuals to trade these currencies and predict the volatility magnitude. To acquire the best results, we used a variety of machine learning approaches and algorithms and compared the models to each other.

Aggressive comments on social media negatively impact human life; Such offensive contents are responsible for depression and suicidal activities. Since online social networking is increasing day by day, the hate content is also getting serious. Hence, several investigations have been published on the domain, such as cyberbullying, cyberaggression, hate speech. We are trying to detect the aggressive comment of three different languages: English, Hindi, and Bangla, using a new behavioral pattern approach. We have also analyzed features such as surface features, linguistic features, sentiment features. Finally, we built a completely new machine-translated dataset and performed the classification using it. The concept of building a machine-translated dataset is to explore a new way of getting a dataset.

Women in Bangladesh are not equally active in access to the technology and banking system. The study focuses on finding out the problems and boundaries that prevent them from using technology and the banking system.

In Bangladesh, the percentage of participation of women in the STEM field is very few. Most importantly, women from rural areas do not get the opportunity to study in the STEM field. We are trying to figure out the boundaries and problems that prevent them from achieving their goal.