Deep Learning for Computer Vision - Lecture 21: Reinforcement Learning

Lecture 21 gives a brief overview of reinforcement learning (RL). We discuss the reinforcement learning problem where an agent interacts with an environment and attempts to maximize a reward signal. We see how RL is more difficult than supervised learning due to challenges such as stochasticity, credit assignment, nondifferentiability, and nonstationarity. We formalize the RL problem as a Markov Decision Process (MDP), and see how value functions and Q-functions can express policies. We discuss value iteration and Q-learning as mechanisms for finding optimal policies, and see that training deep neural networks to approximate Q-functions gives rise to Deep Q-Learning. We also discuss policy gradients as an alternate method to learning optimal policies, and derive the REINFORCE algorithm. Note: This is an extremely brief one-lecture overview of reinforcement learning. This topic could easily fill an entire semester-long course; if you want to learn more I suggest checking out Berkeley CS 285 at http://rail.eecs.berkeley.edu/deeprlcourse/. Slides: http://myumi.ch/mnN5y _________________________________________________________________________________________________ Computer Vision has become ubiquitous in our society, with applications in search, image understanding, apps, mapping, medicine, drones, and self-driving cars. Core to many of these applications are visual recognition tasks such as image classification and object detection. Recent developments in neural network approaches have greatly advanced the performance of these state-of-the-art visual recognition systems. This course is a deep dive into details of neural-network based deep learning methods for computer vision. During this course, students will learn to implement, train and debug their own neural networks and gain a detailed understanding of cutting-edge research in computer vision. We will cover learning algorithms, neural network architectures, and practical engineering tricks for training and fine-tuning networks for visual recognition tasks. Course Website: http://myumi.ch/Bo9Ng Instructor: Justin Johnson http://myumi.ch/QA8Pg