(Neural) CDE & TPCNN-based Regression for Irregularly Sampled Transit Lightcurves

This is still a work in progress, however, this is a neural controlled differential equation and time-parametrised CNN hybrid architecture to extract rotation periods from irregularly sampled time series data such as exoplanet transit lightcurves from TESS. The training data for this model is essentially synthetics produced using `pytansit' (see https://github.com/LoopyNoodle/synthdata-transits). This was a part of my internship project at the Tata Institute of Fundamental Research under Prof. Shravan Hanasoge during the summer of 2025.