Network Analysis and Mining

The main objective of Frequent Subgraph Mining (FSM) is finding subgraph patterns that are frequent across a collection of graphs. Another important task is to find subgraphs which are candidate for motif in a given network. For this, we need to count each topology’s frequency in the input network as well as in many randomized networks. Counting a topology’s frequency in a single network is a challenging task as it requires solving subgraph isomorphism, a known $NP$-complete problem. In both the cases, I propose a sampling based solution which is efficient and works better than state-of-art methods. In [1], I propose a method for frequent subgraph mining, called $FS^{\large 3}$, that is based on sampling of subgraphs of a fixed size (The name $FS^{\large 3}$ should be read as $F-S-Cube$, which is a compressed representation of the 4-gram composed of the bold letters in $\bf F$ixed $\bf S$ize $\bf S$ubgraph $\bf S$ampler. In [2], I propose method on finding concentration of prospective motifs using a novel sampling based method. In subsequent works, I show that the algorithm ($FS^{\large 3}$) I developed for graph mining can be applied to find essential structures when applied to the interfacial region of a set of oligomeric proteins [3] and the algorithm for motif finding [2] can be used to classify apps from google app store as malignant or benign [4].

Publications

[1] Tanay Kumar Saha and Mohammad Al Hasan. “$FS^{\large 3}$: A sampling based method for top‐k frequent subgraph mining.” Statistical Analysis and Data Mining: The ASA Data Science Journal 8.4 (2015): 245-261. [PDF]

[2] Tanay Kumar Saha and Mohammad Al Hasan. “Finding Network Motifs Using MCMC Sampling.” In CompleNet (pp. 13-24). [PDF]

[3] Tanay Kumar Saha, Ataur Katebi, Wajdi Dhifli and Mohammad Al Hasan. “Discovery of Functional Motifs from the Interface Region of Oligomeric Proteins using Frequent Subgraph Mining.” IEEE/ACM Transactions on Computational Biology and Bioinformatics. [PDF]

[4] Wei Peng, Tianchong Gao, Devkishen Sisodia, Tanay Kumar Saha, Feng Li, and Mohammad Al Hasan. “ACTS: Extracting android App topological signature through graphlet sampling.” In 2016 IEEE Conference on Communications and Network Security (CNS), (pp. 37-45). [PDF]