One-Class Learning and Concept Summarization for Vaguely Labeled Data Streams *
Xingquan Zhu, Wei Ding, Philip S. Yu · 2009
In this paper, we formulate a new research problem of concept learning and summarization for one-class data streams. The main objective is to (1) allow users to label instance groups, instead of single instances, as positive samples for learning, and (2) summarize concepts labeled by users over the whole stream. The employment of the batch-labeling raises serious issues for stream-oriented concept learning and summarization, because a labeled instance group may contain non-positive samples and users may change their labeling interests at any time, so the positive samples labeled by users, over the whole stream, may contain multiple concepts. To resolve these issues, we propose a One-Class Learning and Summarization (OCLS) system with two major components. In the first component, we propose a Vague One-Class Learning (VOCL) module for concept learning from data streams by using an ensemble of classifiers with instance and classifier level weighting strategies. In the second component, we propose a One-Class Concept Summarization (OCCS) module which uses clustering techniques and a Markov model to summarize concepts labeled by users, with only one scanning of the stream data. Experimental results on synthetic and real-world data streams demonstrate that the proposed VOCL module significantly outperforms its peers for learning concepts from vaguely labeled stream data. The OCCS module is also able to rebuild a high-level summarization for concepts marked by users over the stream.