Transfer sample-based coupled task learning (TCTL)

A multitask learning method with transfer samples

Simultaneously learning two models in two domains, with the help of transfer samples (reference / corresponding / calibration samples) in both domains.
Typical application: calibration transfer of two devices, or sensor drift correction.
Linear logistic regression and ridge regression under the framework of TCTL were implemented for classification and regression.
ref: K. Yan, and D. Zhang, “Calibration transfer and drift compensation of e-noses via coupled task learning," Sens. Actuators B: Chem., vol. 225, pp. 288-297, Mar., 2016.
Copyright 2015 YAN Ke, Tsinghua Univ. http://yanke23.com, [e-mail]