Michal Pandy
World models are learned approximations of how an environment changes. Imagine a robot arm trying to pick up a mug. If it moves the gripper slightly left, will it make contact? If it closes too early, will the mug slip? A world model is the part that tries to predict these consequences before the robot commits to an action. This is useful, because an agent can evaluate possible actions without…
Energy-Based Models (EBMs) define probability distributions by assigning an energy value to each possible data configuration. Lower energy corresponds to higher probability, which makes EBMs a flexible framework for generative modeling and representation learning. This post introduces the basic formulation, explains why normalization and sampling are difficult, and covers practical training…
In the last five years, I have reviewed more than twenty personal statements that high school students wrote for admission to UK universities. There were many overlapping issues that I thought would be worth aggregating. A small caveat is that I reviewed engineering and natural sciences statements, so take my tips with a grain of salt if you are applying for social sciences. Focus on academics.…