@inproceedings {pub3269,
	title = {Dedicated Memory Models for
Continual Learning in the Presence of Concept Drift},
	author = {Viktor Losing AND Barbara Hammer AND Heiko Wersing},
	year = {2016},
	month = {December},
	abstract = {Learning in non-stationary data streams is is a highly challenging task,
since the fundamentally different types
of possibly occurring drift undermine classical assumptions such as
i.i.d. data or stationary distributions.
Available algorithms are either struggling with certain forms of drift or require
a priori knowledge in terms of a task specific setting.
We propose a method which couples the KNN classifier with dedicated memories for the
current and former concepts and applies them according to
the demands of the given situation. In this way we are able to deal
with heterogeneous concept drift, without the necessity of a task individual hyperparameter
 optimization.
An extensive evaluation on various benchmarks is conducted and provides a detailed picture
of the methods benefits in comparison to state of the art methods.
The highly competitive results throughout all
experiments underline the robustness of the approach as well as its capability
to handle heterogeneous concept drift. },
	publisher = {NIPS Foundation},
	booktitle = {Continual Learning and Deep Networks Workshop NIPS 2016}
}