What it does This repo is a structured six-week course in deep learning, delivered as a pile of PowerPoint presentations and matching code notebooks. It starts at Python basics and MNIST, then marches through CNNs, RNNs, detection, segmentation, reinforcement learning, neural style transfer, and GANs. Each week pairs theory slides with “Let’s implement” exercises.

The interesting bit The curriculum is a time capsule of mid-2010s deep learning enthusiasm — AlexNet and GoogLeNet sit proudly alongside AlphaGo’s MCTS+CNN breakdown, and there’s a whole session on Hangul character RNNs that you won’t find in the usual Western syllabi. The pacing assumes you’ll build a classifier one week and generate adversarial attacks the next.

Key highlights

• Six themed weeks with 30+ presentation files and linked code notebooks
• Coverage spans classic CNNs (AlexNet, ResNet), RNNs/LSTMs, detection (R-CNN family, YOLO), segmentation (FCN, DeepLab), plus GANs and neural style transfer
• Includes less-common topics: visual Q&A, super resolution, Bayesian optimization, adversarial attacks
• Korean-language angle with Hangul char-rnn and handwriting generation
• TensorFlow-era code (pre-PyTorch dominance, judging by the TensorBoard week)

Caveats

• Everything is PowerPoint (.pptx); no rendered slides or web viewer, so you’re downloading blind
• “2nd ed.” implies a first edition existed, but the repo gives no changelog or update date
• Code paths reference master branch naming, suggesting the content has not been refreshed recently

Verdict Worth a look if you’re teaching an intro course and want a proven week-by-week skeleton to adapt, or if you need Korean-language DL materials. Skip it if you want runnable notebooks in a browser or cutting-edge architecture implementations — this is a curated archive, not a living framework.