My name is Alexia and I am a researcher in artificial intelligence at Samsung. In Winter 2019, I started a PhD in Artificial Intelligence at the Montreal Institute for Learning Algorithms (MILA). My background is in both Statistics and Computer Science (BSc in Math and CS, MSc in Statistics). I have expertise in Biostatistics (focused on Gene-by-environment interaction modelling) and Artificial Intelligence (focused on generative modelling).
I’m particularly interested in deep generative models: neural networks that generate any type of data (e.g., images, sound, videos, text, etc.). It is so much more rewarding to generate something than to try to predict something; it also has so much artistic potential. Imagine generating a potential new episode of TNG using only a written script and voice samples! We are extremely far from that goal, but maybe one day slowly getting close to that dream goal!
Some of my previous work include GottaGoFast, Adversarial score matching, Relativistic GANs, the LEGIT model trained using alternating optimization, and a study of the prior of Bayesian Trees. I also generated pictures of cats (Meow Generator) using several types of Generative Adversarial Networks (GAN).