Incremental Learning with Semi-supervised K-means

Frederico Damasceno Bortoloti¹; Patrick Marques Ciarelli¹

¹ Universidade Federal do Espírito Santo

Baixar PDF

Resumo

In many real-world tasks a lot of unlabeled data are collected over time and, although they may be useful to improve the quality of classication models, they are usually ignored. Semi-supervised learning techniques combine unlabeled and labeled data to capture more useful information about a particular task. On the other hand, an incremental learning technique can incorporate new information to an existing model, so that it can dynamically adapt its structure to follow the environment changes. K-means is a well known technique for unsupervised learning, which has adapted versions to supervised learning. K-means works by clustering information providing a node based structure. In order to unify the characteristics of both approaches, in this paper is proposed an incremental semi-supervised learning method called SSK-means, which is based on a supervised k-means method. Experiments were conducted using publicly available benchmark datasets. Results show the proposed method is promising.

Palavras-chave: Semi-supervised learning; Incremental learning; K-means; k nearest neighbors