Towards Safe Weakly Supervised Learning

Abstract

In this paper, we study weakly supervised learning where a large amount of data supervision is not accessible. This includes 4 i) incomplete supervision, where only a small subset of labels is given, such as semi-supervised learning and domain adaptation; ii) 5 inexact supervision, where only coarse-grained labels are given, such as multi-instance learning and iii) inaccurate supervision, where 6 the given labels are not always ground-truth, such as label noise learning. Unlike supervised learning which typically achieves 7 performance improvement with more labeled examples, weakly supervised learning may sometimes even degenerate performance 8 with more weakly supervised data. Such deficiency seriously hinders the deployment of weakly supervised learning to real tasks. It is 9 thus highly desired to study safe weakly supervised learning, which never seriously hurts performance. To this end, we present a 10 generic ensemble learning scheme to derive a safe prediction by integrating multiple weakly supervised learners. We optimize the 11 worst-case performance gain and lead to a maximin optimization. This brings multiple advantages to safe weakly supervised learning. 12 First, for many commonly used convex loss functions in classification and regression, it is guaranteed to derive a safe prediction under 13 a mild condition. Second, prior knowledge related to the weight of the base weakly supervised learners can be flexibly embedded. 14 Third, it can be globally and efficiently addressed by simple convex quadratic or linear program. Finally, it is in an intuitive geometric 15 interpretation with the least square loss. Extensive experiments on various weakly supervised learning tasks, including semi-16 supervised learning, domain adaptation, multi-instance learning and label noise learning demonstrate our effectiveness.

Elevating Prediction Performance for Mechanical Properties of Hot-Rolled Strips by Using Semi-Supervised Regression and Deep Learning

“…To achieve a good model, the maximal performance gain against the baseline regressor can be expressed as an objective function as Eq. (3) [39,40].…”

“…Other parameters are remaining as default. The details of Self-LS and Self-kNN could be found in references [39]. For DNN part, the four hidden layer neural network was used in the investigation.…”

Deep Learning for Change Detection in Remote Sensing Images: Comprehensive Review and Meta-Analysis

“…However, training a change detection framework based on weakly supervised learning (WSL) can alleviate the need for manual annotation. Weakly supervised data include a small quantity of accurate label information, that differs from data in traditional supervised learning [127].…”

A survey of visual analytics techniques for machine learning

“…Weakly supervised learning builds models from data with quality issues, including inaccurate labels, incomplete labels, and inexact labels. Improving data quality can boost the performance of weakly supervised learning models [290]. Most existing methods focus on inaccurate data (e.g., noisy crowdsourced annotations and label errors) quality issues, and interactive labeling related to incomplete data (e.g., none or only a few data are labeled) quality issues.…”