I would like to explain in this post, one the projects I was involved during my Masters degree at ENS Paris Saclay, during the Introduction to Numerical Imaging course.
Introduction InterfaceGAN is a paper that has been published on CVPR 2020, by Yujin Shen et al. It argues that well-trained generative models learns a disentangled latent space representation.
Basically given a generated face image from a random gaussian noise, InterfaceGAN can control some specific attributes of the face without altering the semantic information of it.Why NLP is so exciting? Natural Language Processing is an application of AI and Deep Learning that allows machines and algorithms understand languages (Natural Language) in order to easily deal with any problems related to text (text classification, sentiment analysis, summarization, etc.). There is also a very large interest around NLP from big tech companies and investors as the potential applications of Deep Learning for NLP are becoming more and more impactful.The paper ‘Do pedestrians pay attention? Eye contact detection in the wild’ is finally on arxiv after its submission to the Intelligent Transportation Journal (ITS). You can find the paper, its implementation as well as the released benchmark on this websiteHere is my attempt to explain the notions and intuitions behind Imitation Learning with the best of my knowledge. Credits to this very nice blog where I have learned most of the things that I have understood about the concept, and to this website for the image above. Now let’s directly dive in.
1. Brief intuitions In Reinforcement Learning, you learn to make good sequence of decisionsSource: http://web.stanford.edu/class/cs234/slides/lecture1.pdf
" In Reinforcement Learning, you learn to make good sequence of decisions The field of Reinforcement Learning is an area of machine learning where an intelligent agent interacts with an environment in order to learn a policy (i.We decided with my friend Arthur Zucker to challenge ourselves into this Kaggle competition. The project looks quite exciting, with a serious and challenging application on a real-wolrd problem. According to the official Kaggle repository: A malignant tumor in the brain is a life-threatening condition. Known as glioblastoma, it’s both the most common form of brain cancer in adults and the one with the worst prognosis, with median survival being less than a year.