Research

Projects

Current Projects:

Fairness testing of ML Systems:

Fairness testing of machine learning (ML) systems is a project focused on evaluating and ensuring that ML algorithms are equitable across different demographic groups. This involves analyzing data for biases, assessing the fairness of model outcomes, and implementing strategies to mitigate any detected biases. The objective is to promote transparency, accountability, and inclusivity in ML deployments, ensuring that these technologies do not perpetuate or exacerbate social inequalities.

Ethereum Smart contract testing for Blockchain Systems:

Smart contract testing for blockchain systems involves rigorously assessing smart contracts to ensure they execute as intended without vulnerabilities or errors. This process includes functional testing to verify the correct behavior of the contracts, security testing to identify potential exploits, and performance testing to ensure they handle transactions efficiently under various conditions. Effective testing is crucial to prevent costly errors and enhance the reliability and security of blockchain applications.

Previous Projects:

Quality Assurance of Biomedical Text Processing Tool

Bioinformatics software plays a very important role in making critical decisions within many areas including medicine and health care. We test LingPipe , a tool for processing text using computational linguistics and often used in bioinformatics for extracting biomedical terms from text. Thee typical outputs from a bio-entity recognition process are hard for developers to validate and requires the support from domain experts.  In this project, we are exploring the effectivness of using Metamorphic Testing for testing bioinformatics tool.

Quality Assurance of Automated Protein Function Prediction Tools

Automated Function Prediction (AFP) tools will play a very important role in medicine and health care. The current tools provides different set of output Gene Ontology terms, and only a few terms would be in common with the experimentally validated terms.  As a result, biologists and developers have difficulty in selecting the tools to perform experiment.  In this project, we identify the feasibility of using Metamorphic Testing for testing AFP tools. 

Improving the Effectiveness of Metamorphic Testing

Metamorphic Testing has been successfully applied in different domains and it becomes important to develop techniques that allow for the effective use of MT. The cost (in terms of time and resources) of the MT process is proportional to the number of metamorphic relations (MRs) used for testing. So it becomes important to select a diverse set of MRs which helps to save time and cost during regression testing. We propose novel automated methods for MR prioritization.