Skip to main content

Neural Networks and Deep Learning Study Materials

These materials are designed to serve as a learning path from machine learning fundamentals to deep learning application topics. The first half establishes basic concepts, training workflows, and common models, while the second half progressively enters specialized topics such as convolutional neural networks, recurrent neural networks, computer vision, natural language processing, and graph neural networks.

One advantage of these materials is their clear pacing: each lecture first introduces core concepts, then provides typical tasks or practice examples. This way, you are less likely to stay at the terminology level and more likely to connect models, tasks, and concrete implementations.

  1. 1-Machine Learning Fundamentals
  2. 2-Deep Learning Fundamentals
  3. 3-Convolutional Neural Networks
  4. 4-Recurrent Neural Networks
  5. 5-Advanced Deep Learning Topics
  6. 6-Deep Learning Applications (Computer Vision)
  7. 7-Deep Learning Applications (Natural Language Processing)
  8. 8-Graph Neural Networks

What Each Lecture Covers​

  • 1-Machine Learning Fundamentals: establishes the basic framework for supervised learning and model evaluation, then ties the workflow together through Iris classification.
  • 2-Deep Learning Fundamentals: explains why deep learning is needed, and covers forward propagation, backpropagation, and basic usage of deep learning frameworks.
  • 3-Convolutional Neural Networks: starting from image tasks, introduces the structure, training methods, and classic networks of CNNs.
  • 4-Recurrent Neural Networks: starting from NLP tasks, progressively transitions to word vectors, RNN, LSTM, GRU, and attention mechanisms.
  • 5-Advanced Deep Learning Topics: covers transfer learning, GANs, and reinforcement learning, serving as a bridge from foundational models to more complex topics.
  • 6-Deep Learning Applications (Computer Vision): brings together typical visual tasks beyond classification, including detection, segmentation, tracking, and several specialized tasks.
  • 7-Deep Learning Applications (Natural Language Processing): extends from basic NLP tasks to reading comprehension, question answering, text generation, and multimodal directions.
  • 8-Graph Neural Networks: targets graph-structured data, filling the gap for non-Euclidean data scenarios that conventional CNNs and RNNs cannot directly handle.

How to Use These Materials​

  • If this is your first time systematically studying this area, it is recommended to go in order. At least complete the first four lectures before deciding whether to continue deeper into vision, NLP, or graph learning.
  • If you already know the basic deep learning training workflow, you can treat these materials as a topical guide and skip around based on the tasks you need to work on.
  • Each lecture is best studied alongside code. After reading the concepts, immediately reproduce the practice examples or a minimal runnable example yourself.