1. Developing/Improving Machine Learning models. Based on the learned information, machine learning models are created to carry out specific tasks like classification or regression (training on given data). Around the world, significant efforts are being made to improve the performance of these classifiers and models.
Our team has created a new Bayesian classifier that rivals the performance of well-known algorithms like KNN, SVM, RF, and Naive Bayes. We also work to enhance the effectiveness of currently used techniques by I modifying model parameters and (ii) enhancing the effectiveness of data through pre-processing procedures.
2.Application of Artificial Intelligence/ Machine Learning on real-life problems: In every aspect of our lives, machine learning benefits us. With the incorporation of AI into our daily lives, life has become intelligent. We use machines to help us with difficult and time-consuming activities. We consider if this could be applied to other branches of research and daily life.
We have created a particle picker that automatically chooses protein particles from cryo-EM images (CASSPER). This is a huge step in understanding the structure of proteins, viruses, and other similar entities. This will subsequently aid in the creation of medications and anti-viruses. The newly created tool is on par with the most effective ones now on the field, including TOPAZ, crYOLO, and cryoPICKER. The first method paper on cryoEM structural biology to come out of India was this one.
(Source: Tweet by Director-General of CSIR Prof.Shekar C.Mande)