Diver ML Course Assignments

Description:

This repository host all the notebooks that I wrote during the time I participate in Diver's machine learning course. Which includes:

• Basic ML concepts
• Data analysis
• Re-implementation of most ML methods
• Re-implementation of some Neural Network models
• Re-implementation of some Neural network structures

Final graduation project is a separate repo that can be found here

Curriculum

What is learnt each week? (summary)

• Week 1: Python Introduction
• Week 2: Matrix product and gradient decent
• Week 3: Data Analysis
• Week 4: Classification with sklearn
• Week 7: Machine learning flow, model selection with grid search
• Week 8: Reimplement train-test-split
• Week 9: Reimplement linear regression
• Week 10: Reimplement logistic regression
• Week 11: Reimplement SVM
• Week 12: Reimplement decision tree
• Week 13: Reimplement K-mean
• Week 14: Reimplement ensemble methods (blend/bag/stack)
• Week 15: Reimplement simple neural network
• Week 16: Reimplement general neural network
• Week 17: Reimplement convolution neural network 1D
• Week 18: Reimplement convolution neural network 2D
• Week 19: Introduction to tensorflow
• Week 20: Introduction to keras
• Week 21: Just reading a research paper and write report
• Week 22: Review more research papers
• Week 23: Trying out a model (Faster R-CNN) from some one else's implementation
• Week 24: Image altering to expand image dataset
• Week 25: Try U-net's pre-implementation
• Week 26: Use keras's ResNet-50 and VGG-16 and implement them in a U-net architecture
• Week 27: Introduction to NLP
• Week 28: Reimplement Recurrent Neural Network
• Week 29: Comparison of keras's RNN, LSTM and GRU
• Week 30: Transfer learning with Seq-to-seq model
• Week 31: No code
• Week 32: Final project (different repo)

Technologies Used

• Python
• Numpy
• Sklearn
• Matplotlib
• Pandas
• Seaborn
• Tensorflow
• Keras
• OpenCV
• ...

Screenshots

Interesting graphics

1. Different clustering results from different initial conditions (Week 13)

2. Training result of my 2D CNN (Week 18)

   

3. Kernel trick on SVM (Week 11)

   

4. Gradient decent

   Too slow	Too fast	Good

Setup

Requirements and Installations:

1. Python
2. Some notebooks require specific Python libraries so be warned!

Usage

1. Just clone this repository
2. And run the notebooks
3. [Note] (some notebook need dataset) that's not available here.

Acknowledgements

• Many thanks to the mentors (Diop-san and Cedrick-san) from Diver for supporting me

Contact

Created by @produdez - feel free to contact me or follow my blog on medium ❤️!