28-b LFD: The reinforcement learning setting. скачать в хорошем качестве

28-b LFD: The reinforcement learning setting.

Machine Learning From Data, Rensselaer Fall 2020. Professor Malik Magdon-Ismail touches on aggregation (combining models) and reinforcement learning. Models can be combined during learning, e.g boosting and random forests. Models can also be combined after learning, e.g. bagging and blending. We then touch on reinforcement learning, in particular the defining properties of reinforcement learning and epsilon-greedy strategies trading off exploration with exploitation. We then discuss online decision making and in particular the "interviewing" problem, ending with magdon's 1/e-theorem for dating - a mathematical approach to optimal dating. This is the twenty-eighth lecture in a "theory" course focusing on the foundations of learning, as well as some of the more advanced techniques like support vector machines and neural (deep) networks that are used in practice. Level of the course: Advanced undergraduate, beginning graduate. Knowledge of probability, linear algebra, and calculus is helpful. Some material is from the reinforcement learning e-Chapter of "Learning From Data", amlbook.com, 2012.