TY - GEN
T1 - Approximation algorithms for tensor clustering
AU - Jegelka, Stefanie
AU - Sra, Suvrit
AU - Banerjee, Arindam
N1 - Copyright:
Copyright 2010 Elsevier B.V., All rights reserved.
PY - 2009
Y1 - 2009
N2 - We present the first (to our knowledge) approximation algorithm for tensor clustering - a powerful generalization to basic 1D clustering. Tensors are increasingly common in modern applications dealing with complex heterogeneous data and clustering them is a fundamental tool for data analysis and pattern discovery. Akin to their 1D cousins, common tensor clustering formulations are NP-hard to optimize. But, unlike the 1D case, no approximation algorithms seem to be known. We address this imbalance and build on recent co-clustering work to derive a tensor clustering algorithm with approximation guarantees, allowing metrics and divergences (e.g., Bregman) as objective functions. Therewith, we answer two open questions by Anagnostopoulos et al. (2008). Our analysis yields a constant approximation factor independent of data size; a worst-case example shows this factor to be tight for Euclidean co-clustering. However, empirically the approximation factor is observed to be conservative, so our method can also be used in practice.
AB - We present the first (to our knowledge) approximation algorithm for tensor clustering - a powerful generalization to basic 1D clustering. Tensors are increasingly common in modern applications dealing with complex heterogeneous data and clustering them is a fundamental tool for data analysis and pattern discovery. Akin to their 1D cousins, common tensor clustering formulations are NP-hard to optimize. But, unlike the 1D case, no approximation algorithms seem to be known. We address this imbalance and build on recent co-clustering work to derive a tensor clustering algorithm with approximation guarantees, allowing metrics and divergences (e.g., Bregman) as objective functions. Therewith, we answer two open questions by Anagnostopoulos et al. (2008). Our analysis yields a constant approximation factor independent of data size; a worst-case example shows this factor to be tight for Euclidean co-clustering. However, empirically the approximation factor is observed to be conservative, so our method can also be used in practice.
UR - http://www.scopus.com/inward/record.url?scp=77952042297&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=77952042297&partnerID=8YFLogxK
U2 - 10.1007/978-3-642-04414-4_30
DO - 10.1007/978-3-642-04414-4_30
M3 - Conference contribution
AN - SCOPUS:77952042297
SN - 3642044131
SN - 9783642044137
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 368
EP - 383
BT - Algorithmic Learning Theory - 20th International Conference, ALT 2009, Proceedings
T2 - 20th International Conference on Algorithmic Learning Theory, ALT 2009
Y2 - 3 October 2009 through 5 October 2009
ER -