Active Regression with Adaptive Huber Loss

06/05/2016
by   Jacopo Cavazza, et al.
0

This paper addresses the scalar regression problem through a novel solution to exactly optimize the Huber loss in a general semi-supervised setting, which combines multi-view learning and manifold regularization. We propose a principled algorithm to 1) avoid computationally expensive iterative schemes while 2) adapting the Huber loss threshold in a data-driven fashion and 3) actively balancing the use of labelled data to remove noisy or inconsistent annotations at the training stage. In a wide experimental evaluation, dealing with diverse applications, we assess the superiority of our paradigm which is able to combine robustness towards noise with both strong performance and low computational cost.

READ FULL TEXT
research
07/03/2023

Semi-supervised multi-view concept decomposition

Concept Factorization (CF), as a novel paradigm of representation learni...
research
06/12/2021

Semi-supervised Active Regression

Labelled data often comes at a high cost as it may require recruiting hu...
research
05/18/2022

SemiCurv: Semi-Supervised Curvilinear Structure Segmentation

Recent work on curvilinear structure segmentation has mostly focused on ...
research
01/03/2022

Multi-view Data Classification with a Label-driven Auto-weighted Strategy

Distinguishing the importance of views has proven to be quite helpful fo...
research
06/01/2021

Semi-Supervised Domain Generalization with Stochastic StyleMatch

Most existing research on domain generalization assumes source data gath...
research
03/15/2022

Scalable Penalized Regression for Noise Detection in Learning with Noisy Labels

Noisy training set usually leads to the degradation of generalization an...

Please sign up or login with your details

Forgot password? Click here to reset