Indian universities have always been at the forefront of groundbreaking research in Artificial Intelligence. Highly competent researchers, state-of-the-art facilities and foremost technical know-how give a fertile ground for remarkable AI advances. This article will bring you up to speed on the latest in this field. Following are snippets from our conversations with India’s premier researchers about their recent work on AI-related projects, from deepfake detection to Covid-related work to natural language processing.
- How IIT Ropar created FakeBuster to make video conferences safer: Deepfake technology is being used by perpetrators to fool targets through video calling, as was recently alleged by senior European politicians including from the UK. Dr Abhinav Dhal has developed a tool that singles out impostors attending virtual conferences. Known as FakeBuster, this DL-based solution is a first of its kind and has been tested on popular applications, such as Skype and Zoom. It uses a 3D convolutional neural network (CNN) for predicting video segment-wise fakeness scores.
- How ISI Kolkata researchers developed Deep Learning tool for COVID-19 screening: A team comprising Dr Sushmita Mitra, Dr Nilanjan Saha and Dr Sanhita Basu has focused on using X-rays for Covid-19 screening. This makes the screening process more affordable by replacing CT scans that are expensive. The DETL (Domain Extension transfer learning) algorithm was used in the study.
- How IIIT Hyderabad developed risk stratification algorithms for Covid-19 patients: The CSIR-IIITH-Intel Covid Project is being helmed by professors U. Deva Priyakumar and Vinod PK. It aims to predict mortality of Covid-19 patients through risk stratification algorithms. The idea is to improve the efficiency of the healthcare system by giving targeted care to at-risk patients. The accuracy of their model ranges from between 60% to 90% based on the quality of datasets supplied to the algorithm.
- How IIT Madras is leveraging language technology for India: AI4Bharat is helmed by IIT Madras faculty members Mitesh Khapra and Pratyush Kumar. They aim to improve the digital experience for India’s millions of users who access the web in vernacular languages but are challenged by the extremely low availability of information. They’re focusing on the transliteration problem, in addition to building translation models for Indian languages.
Source: indiaai.gov.in
