Sultan - Research

✅Our Paper: "Towards Addressing Identity Deception in Social Media using Bangla Text-Based Gender Identification" is accepted at the KDD conference.

Sultan

Research

During my M.Sc. thesis, I worked in the natural language processing domain and utilized data science & NLP knowledge to answer research queries in computational linguistics and other real-life problems. As part of my B.Sc. thesis, I have worked on combinatorial optimization algorithms to resolve existing issues of the Simplex Algorithm to maximize profit from limited resources under certain conditions. In addition, as part of different M.Sc. coursework, I have research experience in bioinformatics, systems, machine learning, security, and wireless ad hoc networks.

PhD Research Work

In this work, we explored how deep learning and explainable AI can be used to classify whether distant galaxies are powered by stars or black holes — based solely on their nebular emission lines. The aim is to bridge the gap between observable light and hidden high-energy spectra like EUV, which are often out of reach with traditional telescopes.

This work is being continued with the collaboration of NASA and UMBC.

M.Sc. Thesis

Gender Identification from Social Media Bangla Text Using Recurrent Neural Networks, NLP, Deep Learning: The Gender Identification (GI) problem is a part of text mining concerned with determining the gender of a given text author. This paper uses stylometric and word embedding approaches to identify the author’s gender from Bengali texts. We build deep learning-based models to identify gender from the Bangla text. The traditional machine learning model is used to compare the performance against deep learning models

B.Sc. Thesis

Initial Solution for Simplex Method, Computation Optimization: The Simplex Algorithm can solve linear programming problems in the computation optimization field. However, the cycling problem exists in the simplex algorithm which hinders the algorithm's performance. In this research, we have proposed an algorithm that can remove the cycle problems in the simplex algorithm in linear time. Also, I have started the Simplex algorithm from a particular point to enhance the performance of the algorithm.

Other Research Work

Computational Offloading in Multiple Application-aware Cloudlets

Latency and power-aware offloading are promising issues in mobile cloud computing today. This research presents computation offloading techniques in multiple cloudlets and analyzes them along with their main critical issues. The proposed algorithm is implemented in the mobile cloud computing laboratory of our university and Amazon EC2 server. In the experimental analyses, face recognition and object detection are the tasks to be offloaded to the cloudlets based on the application type.

Supervisor: Dr. Muhammad Abdullah Adnan

Improvement of GPSR Algorithm for MANET

GPSR is a routing protocol for wireless datagram networks that uses the positions of routers and the destination of a packet to make packet forwarding decisions. We are proposing a novel method to increase the performance of the GPSR routing protocol. In this research work, we will compare the performance of GPSR after implementing the method in the GPSR algorithm.

Supervisor: Dr. A.K.M. Ashikur Rahman

An Efficient Approach of Computing Double Cut and Join Distance for Genomes with Duplicate Genes

Genome Rearrangement refers to large-scale mutations in genomes which are responsible for complex changes and structural variations. Computing the edit distance between two genomes is a fundamental problem in the study of genome evolution. The Double Cut and Join computation for two genomes with duplicate genes is NP-Hard. In this study, We propose a new approach to the problem that involves an A* search. We provide a tight heuristic to accelerate the search and also a preprocessing technique to further improve the performance of our solution.

Supervisor: Dr. Md. Shamsuzzoha Bayzid

Performance Research and Optimization on Jython Compared to Python And JAVA

The Python programming language is a descendant of the ABC programming language, which was a teaching language created by Guido van Rossum in the early 1980s. It includes a sizeable standard library and powerful primitive data types. Jython is one of the major interpreters of Python. In this paper, the performance of Jython regarding execution time is evaluated against JAVA and Python concerning the following criteria: memory consumption, CPU utilization, and execution time. Each test was performed in a distributed system on a symmetric multiprocessing system.

Supervisor: Dr. Rifat Shahriyar

Google Sites

Report abuse