This is a repository containing the perfomance analysis notebooks for CEBRA.
- Models notebook is a notebook that documents the search for the best parameters for the CEBRA model to get ideal embeddings for both the circle and the Lorenz attactor
- Lorenz notebook is continuation of the Models notebook that looks more closely at the best parameters to use in obtaining embeddings for the attractor
- PCA_comp is a notebook that compares the output of 2D PCA and the CEBRA embeddings for the Lorenz attractor
- Dimensionality notebook is a notebook used to evaluate the performance of CEBRA models, starting with simple inputs
- d_library has helper functions to be used in the dimensionality analysis of the embeddings
- Animate embeddings is a notebook used to generate animations of the embeddings produced
- Model_arch contains the analysis of the CEBRA embeddings produced by different model architectures
- Temp_analysis is a notebook showing a detailed analysis of how temperature could affect the way the embeddings turn out.