This is an R code to implement Self Supervised Learning (SSL) procedure using the SimCLR (Simple Contrastive Representation). This is a job in progress... As a test dataset, it uses a time series of NDVI data derived from Sentinel-2 (Google Earth Engine) for three trees on a region dominated by Juniper trees in Arizona. The data set is a R file called "NDVI_TimeSeries_Junipers" I am attaching two publications of SSL-SimCLR: The script uses the following libraries: tensorflow, keras, ggplot, recipes, tidyr, dplyr, zoo and purrr. I am attahcing in the respositoy the conda environment containing the requered libraries and python version. The file is called: SimCLR.

Self-Supervised Learning for Time Series_Contrastive or Generative_LIU_2024.pdf

A simple framework for Constrastive Learning of visual representations_CHEN_2020.pdf