I have been working on deep learning for the past couple of months, mostly in deep reinforcement learning. I have recently found an interest in language translation models. There are hundreds of dialects and scripts which will long be forgotten for lack of people using them, these languages, I feel, can be saved/archived through machine translation.
Reasons for my interest in RL is mainly because, it gives an option to apply some of the recent advancement in ML without the need/availability of google/baidu sized data, but by generating data through millions of iterations and in way, it’s always interesting to see how an agent optimizes for a given objective function in RL. I have used traditional machine learning for data analysis and modelling while being a research assistant at an energy research institute during my graduate studies.