SFS Scholars Cohort II

SFS Scholars Cohort II

The Scholarship for Service (SFS) program at Bowie State University, sponsored by the National Science Foundation (NSF), proudly introduces its second cohort of scholars. This prestigious program prepares students for careers in cybersecurity, providing them with financial support, mentorship, and opportunities for cutting-edge research. Faculty mentors, including Dr. Jie Yan, Dr. Avijoy Chakma, Dr. Sarker Kamruzzaman, and Dr. Vivek Mayura Shandilya, lead innovative projects focused on image forensics, scalable machine learning, malware detection, and attack identification. The SFS scholars also have the chance to engage in paid internships with government agencies, equipping them with the hands-on experience and skills needed to address critical cybersecurity challenges and contribute to national security. 

Faculty Research Mentors

Dr Ji Yan

A  professor  in  the  Department  of Computer  Science  and  the  Director  of  Machine Perception   &   Animation   lab   and   Cybersecurity Application  Lab  at  Bowie  State  University.  She  isalso  the  faculty  lead  for  center  for  Cyber  Security  and Emerging   Technologies. Her   areas   of   expertise   include Computer Graphics, Computer Vision, Machine Learning, and Cybersecurity Education. Abstract:  Differentiating real or authentic images from manipulated or fake imagery is significant for image forensics domain and the public. Researchers tend to  develop very complicated deep learning architectures  to  train  detector  models  making  the process of fake image detection expensive to the public and limiting model usability.The goals of  this research project are : 1) to develop a robust forensic analyzer to detect authentic from counterfeit images using deep learning and a less expensive lightweight device with  less  computing  power,  2)  to  produce  a  Raspberry  Pi- based  analyzer  model that will be accessible to the general public,   3)  to   determine   the   best   framework   for   image classification using    raspberry pi, 4)    to evaluate the performance   of   three   frameworks   in   terms   of   image classification  accuracy  and  model  inference  speed  on  IoT raspberry Pi device and, 5) to evaluate the performance of three  types  of  strategies  in  terms  of  average  detection accuracy and model complexity based on the best-performed framework.

Avijoy Chakma

A  incoming   Assistant Professor in the Computer Science department at  Bowie  State  University.  He  completed  his Ph.D.    in    Information    Systems    from    the University  of Maryland, Baltimore County,  in July  2023.  He received a Bachelor of Science degree in Computer Science and    Engineering    from    the    Bangladesh    University    of Engineering and Technology in 2013 and a Master of Science degree in Computer Science from Lamar University, Texas, in 2018. Before his   graduate   study,   he   had   2.5   years   of experience   working   in   the   software   industry.   He   has published multiple peer- reviewed journals and conference papers, and he is the recipient of the Best Paper Award at the IEEE/ACM  CHASE  Conference  2022.  During  his  Ph.D.,  he actively  supervised  multiple  Summer  Research  Experience for  Undergraduates  (REU)  students.  His  research  interests are   cyber-physical   systems,   smart   health,   and   machine learning. Title: Cross-domain Sensing & Modeling for Smart Environments

Abstract:

Cyber-physical  systems   that    create    smart environments are associated with multiple sensing mechanisms  to interact  with  and  understand  the physical world. Efficiently aggregating and processing multi-stream data and developing adaptive machine learning models to make actionable decisions is not trivial.  Often,  smart  sensing (such  as  smartphones, smartwatches,  Google Home,  etc.) device  provided data sources differ significantly from one data source to another. Most existing machine learning approaches need to be more adaptive and scalable to handle such variations. Therefore, developing  scalable  machine  learning  models  under  the presence of multi-stream, multi-modal data sources with labeled and unlabeled information is a challenging avenue to investigate. In this talk, I will discuss the underlying causes of data source   variations,   how   to   overcome   such   data variations to process multiple streams concurrently, and how to develop a scalable, robust machine learning model.

Dr. Sarker Kamruzzaman

Dr.  Sarker’s  primary  research  area  is  Trustworthy  Artificial Intelligence and its applications. His ongoing work seeks to enhance the reliability of AI systems by improving their ability to    explain    decisions    in    simple    terms,    validating    the effectiveness    of    these    explanations,    and    employing knowledge   graphs   for   clearer   insights.   Additionally,   he investigates  AI  applications  in  cybersecurity,  with  recent projects focused on real-time malware detection and secure log-in systems.

Title: Malware detection using Deep Learning

Abstract:

Malware detection is crucial due to the significant threats it poses to both individual systems and networks. Our research explores  the  effectiveness  of  deep  learning  in  adapting  to new threats by leveraging diverse datasets. We employ both traditional    and    advanced    deep    learning    models    for preprocessing, augmenting, training, and detection, working with large datasets such as the SoReL-20M, which contains approximately  20  million  samples.  Our  research  has  led  to improved   model   performance,   enhancing   resilience   and versatility in detecting and classifying malware. Despite these advancements,  the  constant  emergence  of  new  malware requires ongoing efforts to stay ahead of evolving threats.

Dr. Vivek Mayura Shandilya

Title: On Attack graphs and Attack trees in Attack identification

Abstract:

Both attack graphs and attack trees are used as a framework to identify probable attacks using the information from the sensors   detecting   anomalies   in   the   networked   system behaviors.   These   anomalies   could   be   caused   by   both malicious    intrusions    or    internal    malfunctioning    either prompted by external malicious  attacks  or  internal  attacks. A rare case of benign system fault could also be a cause. With the technological improvements in attack graph and attack tree generations, and appropriate computational modeling, here we present a study/ survey of the latest works on both the technologies and the methodologies to present the state of art solutions in the field.

SFS Cohort II