Active learning strategies for semi-supervised DBSCAN

Li, Jundong, Sander, Jörg, Campello, Ricardo, and Zimek, Arthur (2014) Active learning strategies for semi-supervised DBSCAN. In: Lecture Notes in Computer Science (8436) pp. 179-190. From: Canadian AI 2014: 27th Canadian Conference on Artificial Intelligence, 6-9 May 2014, Montréal, Canada.

[img] PDF (Published Version) - Published Version
Restricted to Repository staff only

View at Publisher Website:


The semi-supervised, density-based clustering algorithm SS-DBSCAN extracts clusters of a given dataset from different density levels by using a small set of labeled objects. A critical assumption of SS-DBSCAN is, however, that at least one labeled object for each natural cluster in the dataset is provided. This assumption may be unrealistic when only a very few labeled objects can be provided, for instance due to the cost associated with determining the class label of an object. In this paper, we introduce a novel active learning strategy to select "most representative" objects whose class label should be determined as input for SSDBSCAN. By incorporating a Laplacian Graph Regularizer into a Local Linear Reconstruction method, our proposed algorithm selects objects that can represent the whole data space well. Experiments on synthetic and real datasets show that using the proposed active learning strategy, SSDBSCAN is able to extract more meaningful clusters even when only very few labeled objects are provided.

Item ID: 46782
Item Type: Conference Item (Research - E1)
ISBN: 978-3-319-06482-6
Keywords: active learning, semi-supervised clustering, density-based clustering
Date Deposited: 13 Mar 2017 23:25
FoR Codes: 01 MATHEMATICAL SCIENCES > 0104 Statistics > 010401 Applied Statistics @ 100%
SEO Codes: 97 EXPANDING KNOWLEDGE > 970101 Expanding Knowledge in the Mathematical Sciences @ 100%
Downloads: Total: 4
More Statistics

Actions (Repository Staff Only)

Item Control Page Item Control Page