/AI_Course

Primary LanguageJupyter NotebookMIT LicenseMIT

AI course part1-Machine learning

Project-based learning

Day1

  • using Colab
  • dataset: iris
  • framwork: tensorflow 2.x

What's special?

Develop function: sklearn-based preceptron training visualization

  • using partial_fit
  • support
    • configurable setting batch_size
    • 2D decision boundary visualization for both binary class and multi-class


Day2

  • using Colab
  • dataset1: digits
  • dataset2: fetch_openml--Mnist original
  • models: Logistic regression, multilayer perceptron, SVM
  • framwork: tensorflow 2.x

What's special?

Develop function: Easy functions for visualizing evaluated results during classification tasks

  • using matplotlib
  • support
    • Take a view for both testing image and predicted probabilities(All image)
    • Take a view for both testing image and predicted probabilities(only misclassificated one)




Day3

  • using Colab
  • dataset: kaggle dogs-vs-cats
  • deep learning methods:
    • Fine-turn Vgg16
    • Train by scratch
  • framwork: tensorflow 2.x