Research Experience

• Research Interest

-Pattern Discovery, Association Rule Mining, Intrinsic Dimensionality and Spatial Query Processing.

• PhD Thesis in the field of Pattern Recognition and Data Mining

Under the supervision of Prof. James Bailey, Prof. Lars Kulik and Dr. Lida Rashidi, my thesis topic was "Data Mining for the Individuals"

In my thesis, I demonstrate the significance of personalised patterns over general patterns and the challenges of adapting existing research to the personalised case. I propose a query driven framework to mine objects with subspace similarity. I also develop theoretical foundations to assess the relevance of features relative to the local neighbourhood of the query. Our local analyses enable us to obtain deeper personalised insights about the query as well as quantify the impact of feature combinations on the query characteristics.

• Research Internship in National Institute of Informatics, Tokyo.

Under the supervision of Prof. Michael Houle, my research topic was "Feature Dominance via Intrinsic Dimensionality "

We investigated the importance of having a local measure of feature relevance, relative to the local neighbourhood of a specific query object. A theoretical framework has been developed for the assessment of the relative contributions of features using the local intrinsic dimensionality model.

• Graduate Thesis in the field of Data Mining

Under the supervision of Prof. Md. Rezaul Karim and co-supervision of Dr. Md. Samiullah, my thesis topic was "Fast mining of closed itemsets and lattice structure using DBV and DSBV approach."

In my thesis, I propose new data structures and algorithms for mining closed itemsets and their lattice structure using bit-vector (BV) technique. The proposed approach outperforms the existing techniques in terms of both time and storage because of the efficient manipulation of the dual characteristics: tidset and superset information of an itemset.

• Undergraduate Thesis in the field of Data Mining

Under the supervision of Prof. Chowdhury Farhan Ahmed and co-supervision of Dr. Md. Samiullah, my thesis topic was "Mining minimal multilevel association rules using lattice based approach".

The main concept of my thesis is to determine the association among transaction items from different/cross levels of an itemset hierarchy. My proposed algorithms mine cross-level closed itemsets with their minimal generators and lattice structure from the level-wise closed itemsets by using the parent-child relationship of their lattice structure. The case studies indicate that we mine better quality minimal association rules as compared to the existing techniques.

• Research Collaboration in the field of Spatial Query Processing

Collaborators: Prof. Md. Eunus Ali and Prof. Tanzima from Bangladesh University of Engineering and Technology (BUET), Dhaka, Bangladesh and, Prof. Eggman Tanin and Prof. Lars Kulik from the University of Melbourne, Melbourne, Australia.

Description: We introduce trip planning queries for groups/subgroups and develop strategies to identify Points of Interests (POIs) from each required type (e.g., restaurant, theater) that minimizes the total trip distance.