Week 5: Introduction to Deep Learning in Python

An introduction to deep learning in Python by Dr. Yi-Xin Liu at Fudan University (lyx@fudan.edu.cn).

This is a part of the course: Road to Scientific Research: Powerful Computer Applications (XDSY118019.01).

Lecture date: 2025.10.16

Machine Learning

  • Traditional machine learning: decision trees, random forests, support vector machines, gradient boosting machines, etc.
  • Deep learning: Deep learning: neural networks with multiple layers (e.g., CNN, RNN, Transformer, etc.)

Introduction to Deep Learning

Fundamentals of Deep Learning

Fundamentals of Deep Learning

Fundamentals of Deep Learning

Fundamentals of Deep Learning

Deep Learning Experiments

Do the following experiments in A neural network playgound (http://playground.tensorflow.org/):

  • Choose Gaussian data, use linear activation function to train.
  • Choose Circle data, use linear activation function to train. Does it succeed? How to obtain a successful training?
  • (Optional) Choose Spiral data, find a successful classification neural network model.

Neural Network Architectures

  • MLP or ANN (Multi-Layer Perceptron)
  • CNN (Convolutional Neural Network)
  • RNN (Recurrent Neural Network)
  • Transformer

Deep Learning with Pytorch

Why PyTorch?

PyTorch is currently the most popular deep learning framework not only in Python but all programming languages.

Other options

  • TensorFlow
  • JAX
  • Keras
  • Lux.jl (Julia)

Watch the video for the current state of machine learning frameworks here.

Introduction to PyTorch

Additional Resources for PyTorch

A Simple Walkthrough

Go to 03_python_deep_learning.ipynb.

More Resources for Machine Learning

Why Biological Neurons Are Deep Neural Networks (Youtube Video)

General Machine Learning

LLM

  • Transformer
  • Ollama: open-source machine learning models
  • Popular models: GPT, Gemini, Claude, Deepseek, Qwen