Research‎ > ‎

Past Projects

VALERA: Versatile yet Lightweight Record-and-Replay for Android

This is a completely revamped approach for the record and replay of Android application runs. The novel idea is to record and replay of sensor and network data with minimum overhead, faithfully replay events and inter-app communication among apps using intents. The work will appear on OOPSLA'15.

Automated Patching of Smartphone Software

This ASE'14 work introduced an intuitive technique to automatically patch critical errors before any official patch is released. The core concept of the approach is to learn about the error, identify and categorize the error on the run time and apply a temporary patch to seal the erroneous code off when appropriate failure conditions are met.

RERAN: Record and Replay for Android

I worked on the development of RERAN, a record and replay utility for Android applications. The tool is capable of recording user interactions regarding sensor events and provides functionality to replay the actions later. RERAN solves particular problems unseen in previous replay tools for Android. For example, it allows replaying gestures such as swipes, pinch, and zoom. Moreover, it is agnostic over the application binaries. Either it is written in Dalvik or native code, RERAN can carry out the replay.

To learn more about RERAN look at here.
The work was published in ICSE 2013.

A Thread Management Annotation Framework for Effective Resource Usage

I have devised an annotation framework for C++ to manage multi-threaded programs. My goal was to run a multi-threaded application in a more efficient way to optimize energy usage.
You can download/view the paper here.
You can also view/download the source code.

Undergraduate Thesis: Detection of Network Building Blocks in Biological Networks

My thesis topic was “Algorithms for Finding Network Building Blocks in Biological Networks.” The goal of this research was to search complex biological networks and find out the network motifs; the non-overlapping patterns (sub-graphs) that recur within a network much more often than expected at random. I proposed an idea to transform the network (e.g., protein-protein interaction network) into a graph, and discover the edges in the network which create the path among the motifs. I devised an algorithm based on Maximum Flow-Minimum Cut theorem. I also wrote a C++ program on that algorithm. My program was tested on datasets from Human Protein Reference Database, The results were very promising. I am working on the expansion of this problem with my partners till now. Especially, I am experimenting on the performance of existing algorithms regarding Maximum Flow-Minimum Cut problem, which is the core concept in our proposed algorithm.

You can find my complete paper here. Also, my code using the Maximum Flow-Minimum cut technique is included. Some of my presentation slides are also given.

My Undergraduate Thesis: View    Download
Maximum Flow-Minimum Cut algorithm for non-overlapping network motif detection: C++ Code
Thesis Presentation 1: View    Download
Thesis Presentation 2: View    Download

DCJ Path Formulation for Genome Transformation

I carried out an independent study on the topic of DCJ (Double cut and join operation) path formulation for Gnome Transformation. My goal was to describe algorithms on the genome rearrangements with matched or unmatched pair of genes in a genome having different gene content. I studied several approaches regarding the transformation of one genomic arrangement to a target genome structure. I presented my research result with an outline of existing ideas at the end of the course. Apart from the research work, I have explored various problems in computational biology such as computing variants of the longest common subsequence, shortest superstring, motif finding, median string finding, and sorting by reversals problems.

You can have my initial presentation on the topic here.

MS Bio-Informatics Study Group Presentation 1: View    Download

tumblr site counter