Research Interests

 Deep Generative Models: Generative modeling aims to learn unknown, complicated, high dimensional probability distribution that data samples belong to. By means of deep generative models, one can obtain the likelihood of an arbitrary image to investigate whether it comes from the same distribution as data samples come and further generate new samples from the underlying data distribution.

 Representation Learning:  I am also interested in the extraction of high-level, abstract and robust representations that can later be used for downstream tasks using transfer learning. Recently I have focused on Diffusion Based Representation Learning. Specifically, I have concentrated on Diffusion Based Autoencoders to learn decodable representations.

Probabilistic Inference:  Probabilistic Inference uses Probabilistic Models as an attempt to describe statistical problems in terms of probability theory and probability distributions.