murat-u-saglam/l_m_l

Learning Machine Learning

The aim is to be exposed to ML models using PyTorch and experiment with basic implementations.

Directory Structure

• Supervised
  • NumerAI
    • Is a hedgefund that supplies features and targets to train models on
    • Tried this since it is a larger dataset and I get to experiment with trees and also build data pipelines
  • Dogs-Cats
    • Basic CNN - Tried this to get exposure for images and classification
• Unsupervised
  • HeartAttack
    • For more data analysis purposes and clustering
• Reinforcement
  • Not implemented yet
• APP
  • Cifar10
    • Creating a basic web app to get images and classify them
  • RealtimeObjectDetection
    • Self-explanatory

Future developments

• Make all models work with GPU training
• Get 90> Accuracy on CNN
• Do some more investigatory stuff on the clustering
• Build reinforcement learning section
• Potentially in different repo - Work on NumerAI more thoroughly

Learn the theory

PCA