Content

Korean Version

• Deep Reinforcement Learning

  I highly recommend the Deep Reinforcement Learning lectures by Prof. Sung Kim, available on the
  Everyone’s Deep Learning website.
  The lectures are in Korean, each about 15 minutes long, and can be completed in a single day.
  The series starts from the basics of Q-learning and builds up to DQN, explaining the motivation and algorithmic workflow in a very intuitive way.

  These lectures were extremely helpful for building a solid foundation in reinforcement learning.

• Auto Encoder

  Dr. Hwal Seok Lee’s video series, Everything About Autoencoders
  (YouTube link), provides a detailed, 3-part introduction to Autoencoders, one of the fundamental generative models.

  Since the lectures include some mathematical formulations early on, I recommend them to those who already have basic knowledge of deep learning and generative models. They were very helpful in understanding the overall structure and intuition behind autoencoders.

English Version

• Deep Unsupervised Learning

  I strongly recommend UC Berkeley Prof. Pieter Abbeel’s
  Deep Unsupervised Learning course
  (course website).
  You can watch the lectures via this
  YouTube playlist.
  It is a gem of a course that covers a wide range of generative models and provides insights into recent trends in the field.

• Deep Learning in Computer Vision

  A widely acclaimed course from Stanford University,
  CS231n: Convolutional Neural Networks for Visual Recognition, offers comprehensive material on applying deep learning to computer vision.
  All resources are available on the official website:
  https://cs231n.stanford.edu/
  The course provides excellent intuition and practical understanding of CV workflows. Since the programming assignments are publicly available, I highly recommend trying them out in Colab for hands-on experience.