Professor Adil Khan

Adil Khan is a Professor of Machine Learning (ML) and Artificial Intelligence (AI). He has more than sixteen years of research, development, and teaching experience in AI and ML. His work comprises both traditional machine learning methods and deep learning techniques.

He has developed industrial solutions for Action and Expression Recognition, Remote Sensing, Medical Image Analysis, Natural Language Processing, Crime Detection, and Accident Detection problems. As for theoretical research, his work aims to help ML find answers to some of the most critical questions. For example, how to train ML models in the absence of large amounts of training data? How to improve the generalization of deep neural networks? How to enable ML models to adapt and generalize to new target domains? How to protect such models from adversarial attacks? How to ensure that these models would make fair decisions? What causes catastrophic forgetting in deep neural networks, and how can we overcome it?

Exemplar Projects:

Attention Sparsity for Efficient Document Ranking, 2022 - 2023

Goal: Designing novel sparse attention mechanisms to improve the efficiency and relevancy of the search engine result page (SERP).

Role: Principal Investigator, Amount: $150K, Funding Body: Huawei

Multi-modal Deep Rank for Efficient Document Ranking, 2021 - 2022

Goal: Designing deep ranking networks to improve the relevancy of the search engine result page (SERP) via incorporating alternative modalities into the Semantic Vector Space.

Role: Principal Investigator, Amount: $150K, Funding Body: Huawei

Fair, Robust and Life-long Machine Learning, 2021 - 2023

Goal: Designing new representation learning techniques to ensure fair and robust machine learning models that are capable of incremental learning without falling victim to catastrophic forgetting.

Role: Lead Scientist, Amount: $10 Million, Funding Body: The Analytical Center of Artificial Intelligence of Innopolis University

Robust Data Augmentation for Deep Networks, 2019 - 2020

Goal: Designing and building a new data augmentation methods to fine-tune trained deep neural networks to improve their generalization performance for financial markets.

Role: Principal Investigator, Amount: $100K, Funding Body: Sermaya Financial