Hire an Illini
Adit Krishnan
- Advisor:
- Hari Sundaram
- Departments:
- Areas of Expertise:
- Graph Mining, User Modeling, NLP
- Recommendation Systems, Info Retrieval
- Deep Learning and Neural Networks
- Thesis Title:
- User-Centric Deep Learning Methods and Applications
- Thesis abstract:
- My research broadly encompasses the development of user-centric deep learning and machine learning models, algorithms and frameworks across several applications - Recommendation Systems, Information Retrieval, Graph Mining, and User Behavior Modeling. The goal of my research is to enable us to adapt neural models to our performance objectives under real-world data and resource constraints, beyond just the overall performance metrics. We aim to identify implicit assumptions that underlie commonly employed training algorithms and model designs, and to address them in an architecture-agnostic manner with solutions that extend to whole classes of models, such as graph convolution networks and neural collaborative filtering. Towards these objectives, I have developed learning approaches encompassing adversarial training, meta learning, transfer learning and adaptive regularization.
- Downloads:
Contact information:
aditk2@illinois.edu