Lecture 29

• Lecture 29
  • Q-learning
  • Inverse RL
  • Supervised Learning

Q-learning

• come up with reward and policy
• take seq of decision
• so that it is maximized by Q function
• reward depends upon game

Inverse RL

• learning reward by observing an agent who is doing things in right way
• and ek bar reward a gaya, tab policy nikal sakta

Supervised Learning

• student ka ht,wt,age tuple vs time to run 100m ka data hai
• given a new student classify how good he is
• this is supervised
• classification
• regression
• trying to learn a function to do classification or regression