Setup

• install conda: https://docs.conda.io/projects/conda/en/latest/user-guide/install/download.html
• clone the repositority

iris dataset

• create a conda environment with required packages (in console): conda create -n ml_workshop scikit-learn matplotlib ipykernel

• Activate conda environment and start the Jupyter notebook:

    conda activate ml_workshop
    python -m ipykernel install --user --name ml_workshop --display-name "ML Workshop"
    jupyter notebook
  

• Alternative: run in VSCode with Python and Jupyter notebook extensions

• if you don't want to set up a local version, you can try this browser-based notebook: https://jupyter.org/try-jupyter/retro/notebooks/?path=notebooks/Intro.ipynb

Dogs vs. Cats dataset

Dataset is available on Kaggle - download and extract the train folder into your working directory.

• create a conda environment with required packages (in console): conda create -n ml_workshop2 tensorflow-gpu This requires your GPU to be CUDA-enabled - you may have to install additional drivers depending on your hardware.

• alternative: train using CPU (much slower): conda create -n ml_workshop2 tensorflow

• Install remaining libraries:

    conda activate ml_workshop2
    conda install keras scikit-learn ipykernel pillow h5py
    python -m ipykernel install --user --name ml_workshop2 --display-name "ML Workshop - cats vs. dogs"
    jupyter notebook