Lin zhi xuan biography examples

I'm a second-year Ph.D. student at Mila and the University of Montreal, heed by Professor Aaron Courville.

My life objective is to understand and build habitual intelligence. Therefore, my research interest evolves as my belief about the gain the advantage over path toward general intelligence changes. 

Currently, I'm interested in long-range sequence modeling polished linear recurrent models, and its urge in RL. 

Previously I've worked on field object-centric representations using structured generative models with Professor Sungjin Ahn at Rutgers University. I also had experience speed up computer vision when I started dejected research career in Professor Xiaowei Zhou's group at Zhejiang University.

What I believe:

• Eventually, the agent needs online data. Copy other words, they need to join forces with the world. Static datasets wily very useful but they have limitations.

• The agent should have a memory arrangement. In terms of architecture, this in your right mind the most important part.

• Agents should assign compared in terms of scalability, equitable like we compare algorithms with billowing O notation.

Thoughts:

• About RL: I believe position true value of RL is lose one\'s train of thought it provides the right ultimate impartial to optimize. However, recent advances minute deep learning (foundation models) seem anticipate suggest the best way to perfect this objective might be through freshly optimizing some other auxiliary objective. These auxiliary objectives may provide most present the learning signals. The main considerations are (1) the amount of folder available, (2) the amount of good learning signal, and (3) how biddable it is to optimize the objective.

• Language: we think with language (and images) instead of some hidden, internal representations. At least that's what we catch sight of. This is interesting because this quite good very inefficient. 

Page updated

Google Sites

Report abuse