Youngjae Min

Great work in collaboration with @KwangjunA and @NavidAzizan 3/3

ORFit also establishes a connection between two seemingly distinct algorithms in continual learning and adaptive filtering, namely orthogonal gradient descent (OGD) and recursive least-squares (RLS). Check out the paper here: arxiv.org/abs/2207.13853 Accepted in IEEE CDC'22 2/3

Can we learn sequentially available data without retraining on previous datapoints? We propose 𝗢𝗥𝗙𝗶𝘁 (Orthogonal Recursive Fitting), an algorithm for "one-pass" learning which seeks to fit every new datapoint while minimally changing the predictions on previous data. 1/3