Targeted Adversarial Discriminative Domain Adaptation
TADDA stands for Targeted Adversarial Discriminative Domain Adaptation. TADDA is a software package that enables building Aided Target Recognition (AiTR) and other algorithms for environments and targets where the data/labeled data is scarce by performing semi-supervised domain adaptation. It is applicable to many different domain adaptation scenarios, including from one imaging condition to another, from one imaging modality to another, from simulated data to collected data, and from a set of target classes to a different set of classes.
METHOD: GAN, semi-supervised learning
KEYWORDS: Machine Learning, Deep Learning, Domain Adaptation
TECHNICAL SPECS: Python,
RESEARCH: Targeted adversarial discriminative domain adaptation. Journal of Applied Remote Sensing, 15(3), p.038504, 2021