The nutcracker framework for ensemble interpretability

O. Zeev Ben Mordehay, W. Duivesteijn, M. Pechenizkiy

Research output: Contribution to conferencePosterAcademic

56 Downloads (Pure)

Abstract

The basic principles behind ensembles (e.g.
Random Forest, AdaBoost) are simple. But we’re still
in trouble when attempting to explain the logic taken.
Where does the problem lie? The reason that ensembles
are effective is that the base estimators "work
together" and compensate each for the others’
shortcomings.
The Nutcracker Framework Given a trained ensemble
and the relevant training / test dataset, construct prediction
matrix, M, cases (rows) against predictions (columns).
Bicluster M to a given number of R x C biclusters.
Now, investigate performance per bicluster (R x C).
Identify feature importance per base estimators group (C).
Describe each of the R cases subgroups in terms of
features and values. We use Exceptional Model Mining for that task.
Performance of the ensemble against the dataset
compared to performance of base estimator groups
against subgroups of cases, adds transparency.
Original languageEnglish
Publication statusPublished - Oct 2017
Event16th International Symposium on Intelligent Data Analysis (IDA 2017), October 26-28, 2017, London, UK - London, United Kingdom
Duration: 26 Oct 201728 Oct 2017
http://www.dcs.bbk.ac.uk/ida2017/

Conference

Conference16th International Symposium on Intelligent Data Analysis (IDA 2017), October 26-28, 2017, London, UK
Abbreviated titleIDA 2017
CountryUnited Kingdom
CityLondon
Period26/10/1728/10/17
Internet address

Fingerprint Dive into the research topics of 'The nutcracker framework for ensemble interpretability'. Together they form a unique fingerprint.

  • Cite this

    Zeev Ben Mordehay, O., Duivesteijn, W., & Pechenizkiy, M. (2017). The nutcracker framework for ensemble interpretability. Poster session presented at 16th International Symposium on Intelligent Data Analysis (IDA 2017), October 26-28, 2017, London, UK, London, United Kingdom.