Deep Learning · Robustness · Efficiency
I study how neural networks learn, generalise, and scale.
I'm Dhananjay Tomar, a deep-learning researcher working on robust representation learning and efficient model design. I study how models can generalise across domains, process inputs too large for GPU memory, and remove redundant computation—often through challenging problems in computer vision and computational pathology.
Featured Research
Nuclear masks for robust histopathology
Models trained on data from one hospital often struggle when used on data from other hospitals. We guide the training of a standard image classifier using nuclear masks, helping it focus more on nuclear shape and arrangement. This approach improves accuracy and robustness across hospitals, and the model does not need masks during testing.
View project →Latest Writing
What happens when a cancer model can see only cell nuclei?
The story behind our NeurIPS 2024 paper on nuclear shape, hospital shifts, and robust cancer classification.
Read →
4 deep learning papers that will surprise you
Some lesser known papers which show some amazing aspects of neural networks you didn't know of.
Read →