theovidal/ml-tests

🧠👨‍🎓 Machine Learning models developed to learn these techniques

Miscellaneous Machine Learning code

Little projects I code to learn machine learning algorithms and data science processes.

What I developed

In chronological order, starting from July 2024:

• Simple linear regression
• IMDB sentiment classification with MLP and CNN
• A Kaggle classification competition
• Weather forecast with timeseries and RNN
• Cats and Dogs image classification with CNN, with a deep dive into what the model learns (printing the patterns)
• [WIP] Translation with a Transformer model
• Hi! Paris Data Bootcamp: a one-week training on data cleaning, visualization, ML and explainability, with ethics considerations and business cases

What I learned

Using Numpy, Pandas and TensorFlow (with Keras):

• Multi-layer Perceptron (MLP)
• Convolutional Neural Network (CNN)
• Recurrent Neural Network (RNN) with Long short-term memory (LSTM)
• Autoencoders
• Attention blocks and Transformer architecture
• Cleaning and visualizing structured data, text, images, timeseries
• Explainability of a model

I compiled all theoretical notions I learned on a single Notion page (in French only for the moment).

My projects around ML

• Beta project: using AI for rock climbing, for instance predicting route grades or generating new routes
• Go Neural: implementing a neural network from scratch in Go

Credits

Here are all the resources I used to code these projects:

• TensorFlow documentation
• F. Chollet - Deep Learning with Python