TY - JOUR
T1 - Robust semi-supervised least squares classification by implicit constraints
AU - Krijthe, Jesse
AU - Loog, Marco
PY - 2017
Y1 - 2017
N2 - We introduce the implicitly constrained least squares (ICLS) classifier, a novel semi-supervised version of the least squares classifier. This classifier minimizes the squared loss on the labeled data among the set of parameters implied by all possible labelings of the unlabeled data. Unlike other discriminative semi-supervised methods, this approach does not introduce explicit additional assumptions into the objective function, but leverages implicit assumptions already present in the choice of the supervised least squares classifier. This method can be formulated as a quadratic programming problem and its solution can be found using a simple gradient descent procedure. We prove that, in a limited 1-dimensional setting, this approach never leads to performance worse than the supervised classifier. Experimental results show that also in the general multidimensional case performance improvements can be expected, both in terms of the squared loss that is intrinsic to the classifier and in terms of the expected classification error.
AB - We introduce the implicitly constrained least squares (ICLS) classifier, a novel semi-supervised version of the least squares classifier. This classifier minimizes the squared loss on the labeled data among the set of parameters implied by all possible labelings of the unlabeled data. Unlike other discriminative semi-supervised methods, this approach does not introduce explicit additional assumptions into the objective function, but leverages implicit assumptions already present in the choice of the supervised least squares classifier. This method can be formulated as a quadratic programming problem and its solution can be found using a simple gradient descent procedure. We prove that, in a limited 1-dimensional setting, this approach never leads to performance worse than the supervised classifier. Experimental results show that also in the general multidimensional case performance improvements can be expected, both in terms of the squared loss that is intrinsic to the classifier and in terms of the expected classification error.
KW - Least squares classification
KW - Robust
KW - Semi-supervised learning
UR - http://www.scopus.com/inward/record.url?scp=84998774649&partnerID=8YFLogxK
U2 - 10.1016/j.patcog.2016.09.009
DO - 10.1016/j.patcog.2016.09.009
M3 - Article
AN - SCOPUS:84998774649
SN - 0031-3203
VL - 63
SP - 115
EP - 126
JO - Pattern Recognition
JF - Pattern Recognition
ER -