Dip-based Deep Embedded Clustering with k-Estimation

Dip-based Deep Embedded Clustering with k-Estimation

Abstract

The combination of clustering with Deep Learning has gained much attention in recent years. Unsupervised neural networks like autoencoders can autonomously learn the essential structures in a data set. This idea can be combined with clustering objectives to learn relevant features automatically. Unfortunately, they are often based on a k-means framework, from which they inherit various assumptions, like spherical-shaped clusters. Another assumption, also found in approaches outside the k-means-family, is knowing the number of clusters a-priori. In this paper, we present the novel clustering algorithm DipDECK, which can estimate the number of clusters simultaneously to improving a Deep Learning-based clustering objective. Additionally, we can cluster complex data sets without assuming only spherically shaped clusters. Our algorithm works by heavily overestimating the number of clusters in the embedded space of an autoencoder and, based on Hartigan's Dip-test - a statistical test for unimodality - analyses the resulting micro-clusters to determine which to merge. We show in extensive experiments the various benefits of our method: (1) we achieve competitive results while learning the clustering-friendly representation and number of clusters simultaneously; (2) our method is robust regarding parameters, stable in performance, and allows for more flexibility in the cluster shape; (3) we outperform relevant competitors in the estimation of the number of clusters.

Grafik Top
Authors
  • Bauer, Lena G. M.
  • Leiber, Collin
  • Schelling, Benjamin
  • Böhm, Christian
  • Plant, Claudia
Grafik Top
Shortfacts
Category
Paper in Conference Proceedings or in Workshop Proceedings (Paper)
Event Title
27th ACM SIGKDD Conference on Knowledge Discovery & Data Mining (KDD)
Divisions
Data Mining and Machine Learning
Event Location
Singapore
Event Type
Conference
Event Dates
14.-18.08.2021
Series Name
KDD '21: Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery & Data Mining
ISSN/ISBN
978-1-4503-8332-5
Page Range
pp. 903-913
Date
14 August 2021
Export
Grafik Top