Adaptive Active Classification of Cell Assay Images

  • Nicolas Cebron
  • Michael R. Berthold
Conference paper

DOI: 10.1007/11871637_12

Part of the Lecture Notes in Computer Science book series (LNCS, volume 4213)
Cite this paper as:
Cebron N., Berthold M.R. (2006) Adaptive Active Classification of Cell Assay Images. In: Fürnkranz J., Scheffer T., Spiliopoulou M. (eds) Knowledge Discovery in Databases: PKDD 2006. PKDD 2006. Lecture Notes in Computer Science, vol 4213. Springer, Berlin, Heidelberg

Abstract

Classifying large datasets without any a-priori information poses a problem in many tasks. Especially in the field of bioinformatics, often huge unlabeled datasets have to be explored mostly manually by a biology expert. In this work we consider an application that is motivated by the development of high-throughput microscope screening cameras. These devices are able to produce hundreds of thousands of images per day. We propose a new adaptive active classification scheme which establishes ties between the two opposing concepts of unsupervised clustering of the underlying data and the supervised task of classification. Based on Fuzzy c-means clustering and Learning Vector Quantization, the scheme allows for an initial clustering of large datasets and subsequently for the adjustment of the classification based on a small number of carefully chosen examples. Motivated by the concept of active learning, the learner tries to query the most informative examples in the learning process and therefore keeps the costs for supervision at a low level. We compare our approach to Learning Vector Quantization with random selection and Support Vector Machines with Active Learning on several datasets.

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

Copyright information

© Springer-Verlag Berlin Heidelberg 2006

Authors and Affiliations

  • Nicolas Cebron
    • 1
  • Michael R. Berthold
    • 1
  1. 1.ALTANA Chair for Bioinformatics and Information Mining, Department of Computer and Information ScienceUniversity of KonstanzKonstanzGermany

Personalised recommendations