Student ( Fall 2011 - present )
Assured Information Management and Sharing (AIMS) Lab
Mathematics & Computer Science
Address: 400 Dowman Dr. W401, Atlanta, GA 30322,
Office: Math and Science Center N414, Emory University
Email: Layla.Pournajaf AT mathcs.emory.edu
I am a PhD student in Computer Science and Informatics program at Emory University (working under supervision of Prof. Li Xiong and Prof. Vaidy Sunderam). I received a bachelor’s degree in Computer Engineering in 2006 and a master's degree in Information Security in 2009 both from Amirkabir University (Tehran Polytechnic) in Iran. My research interests include spatio-temporal data analytics, data privacy, and data mining.
Next location prediction using GPS trajectories
Research Question: How to predict the next location of individuals using their partial trajectories?
Methods: Sequential Pattern Mining, Markov Chain Models, Density-based Clustering
Tools: Python, Java
Data: GeoLife GPS Trajectories
Finding new strategies to reduce maternal mortality
Research Question: How to reduce maternal mortality rate in Mexico?
Methods: Random Forests, Recursive Feature Selection, SVM Classification
Tools: Python, R, IPython Notebook, PostgreSQL, QGIS, Tableau
Git-hub: Data Science For Social Good: Mexico
Accurate bus arrival time prediction using real-world data
Research Question: How to reduce the waiting time in bus stops in highly populated cities like Atlanta? How to save the valuable time of people on daily basis using the existing real-time data?
Methods: Auto-regression Model, Decision Trees, Time series Analysis
Tools: Python, R, PostgreSQL
Data: MARTA (Metropolitan Atlanta Rapid Transit Authority)
Mobile crowd sensing task assignment with uncertain trajectories
♦ This project is a part of a larger project PREDICT: PRivacy Enhancing Dynamic Information Collection and moniToring which aims to build a holistic framework for PRivacy and security Enhancing Dynamic Information Collection and moniToring using feedback loops.
Research Question: How to recruit the best set of participants of a crowd sensing application (e.g. traffic monitoring, urban planing, noise or pollution studies) without knowing the individual's exact locations (e.g. to protect individual’s privacy)?
Methods: Markov Model, Bayesian Networks, Kalman Filtering, Uncertainty Modeling, Linear programming
Tools: Java, R, Matlab
Data: Gowalla Dataset