NASA ARSET: Data Loaders for Training ML Models on Irregularly-Spaced Time-Series Imagery, Part 2/3

Large Scale Applications of Machine Learning using Remote Sensing for Building Agriculture Solutions Part 2: Data Loaders for Training ML Models on Irregularly-Spaced Time-Series of Imagery Trainers: Sean McCartney Guest Instructors: John Just (Deere & Co.), Erik Sorensen (Deere & Co.) -Follow the process to set up a Tensorflow data loader that works with -Parquet files to create a training pipeline suitable for training a model on large-scale data -Perform steps to manipulate the imagery data stored in tables, normalize the values, and bucketize irregularly spaced time-series data to prep for modeling -Follow steps to parallelize/prefetch preprocessing for fast training -Apply the correct procedure to split time-series data into train/val/test sets to avoid data leakage You can access all training materials from this webinar series on the training webpage: This training was created by NASA’s Applied Remote Sensing Training Program (ARSET). ARSET is a part of NASA’s Applied Science’s Capacity Building Program. Learn more about ARSET: