Learning From a Complementary-Label Source Domain: Theory and Algorithms

Transition from Kinetic Theory to Macroscopic Fluid Equations: a Problem for Domain Decomposition and a Source for New Algorithms

In the paper we discuss the transition from kinetic theory to macroscopic uid equations, where the macroscopic equations are deened as asymptotic limits of a kinetic equation. This relation can be used to derive computationally eecient domain decomposition schemes for the simulation of rareeed gas ows close to the continuum limit. Moreover, we present some basic ideas for the derivation of kine...

Domain Adaptation for Learning from Label Proportions Using Self-Training

Learning from Label Proportions (LLP) is a machine learning problem in which the training data consist of bags of instances, and only the class label distribution for each bag is known. In some domains label proportions are readily available; for example, by grouping social media users by location, one can use census statistics to build a classifier for user demographics. However, label proport...

UNDERSTANDING MACHINE LEARNING From Theory to Algorithms

Active Learning Algorithms for Multi-label Data

The iterative supervised learning setting, in which learning algorithms can actively query an oracle for labels, e.g. a human annotator that understands the nature of the problem, is called active learning. As the learner is allowed to interactively choose the data from which it learns, it is expected that the learner would perform better with less training. The active learning approach is appr...

Learning from Complementary Labels

Collecting labeled data is costly and thus is a critical bottleneck in real-world classification tasks. To mitigate the problem, we consider a complementary label, which specifies a class that a pattern does not belong to. Collecting complementary labels would be less laborious than ordinary labels since users do not have to carefully choose the correct class from many candidate classes. Howeve...