Siddharth, who is part of the Department of Engineering’s Computational and Biological Learning Lab, will use the $15,000 awarded to him, to complete research during the academic year 2020–21 as part of his doctoral thesis titled Continual Learning and Federated Learning with Neural Networks.
Siddharth’s research interests are in designing algorithms for large-scale machine learning systems that learn sequentially (called continual learning), without revisiting past data (because of computational or privacy constraints), and for systems that split data among many different clients (called federated learning), data which cannot be shared (such as private data on mobile phones).
“Current machine learning systems achieve great performance by training offline on a dataset. This assumes that data is available in one place and can be trained on as many times as required. My research considers two scenarios that are more representative of many real-world applications,” he said. “Current approaches to handle such real-world constraints tend to be expensive and inaccurate. My research looks at designing new scalable algorithms for these two fields (continual learning and federated learning), exploring and taking advantage of the similar technical challenges faced in both. I am applying Bayesian deep-learning methods to do so. These methods have the potential for good calibration for decision-making and uncertainty propagation for continual learning.”
“I’m delighted to be a recipient of this award,” he added. “It is great to receive recognition that my research has the potential for real-world impact in the future. I will use some of the award money to fund my PhD for longer, allowing me to work on the proposed research before completing my thesis. Secondly, I will visit research groups that I collaborate closely with, as visits such as these are crucial in accelerating research, and it will also allow me to personally develop as a researcher.”
The Awards, in their inaugural year, recognise outstanding doctoral students in computing-related fields in Europe, the Middle East, Africa, and Latin America, who are doing important research at the forefront of technology.
Siddharth will be invited to attend a virtual PhD Summit to meet fellow doctoral students and Microsoft researchers in his field (relating to data artificial intelligence (AI), confidential computing and healthcare intelligence), as well as receiving career coaching, and the opportunity to present his research to a Microsoft Research audience.
Reproduced courtesy of University of Cambridge, Department of Engineering