Publication detail

Obtaining word embedding from existing classification model

ŠŮSTEK, M. ZBOŘIL, F.

Original Title

Obtaining word embedding from existing classification model

Type

conference paper

Language

English

Original Abstract

This paper introduces a new technique to inspect relations between classes in classification model. The method is built on the assumption that it is easier to distinguish some classes than others. The harder the distinction is, the more similar the objects are. Simple application demonstrating this approach was implemented and obtained class representations in a vector space are discussed. Created representation can be treated as word embedding where the words are represented by the classes. As an addition, potential usages and characteristics are discussed including a knowledge base.

Keywords

unsupervised learning, artificial intelligence, word embedding, word2vec, CNN

Authors

ŠŮSTEK, M.; ZBOŘIL, F.

Released

22. 3. 2018

Publisher

Springer International Publishing

Location

Cham

ISBN

978-3-319-76347-7

Book

Intelligent Systems Design and Applications

Edition

ISDA 2017 Intelligent Systems Design and Applications

ISBN

2194-5357

Periodical

Advances in Intelligent Systems and Computing

Year of study

2018

Number

736

State

Swiss Confederation

Pages from

540

Pages to

547

Pages count

8

URL

BibTex

@inproceedings{BUT147178,
  author="Martin {Šůstek} and František {Zbořil}",
  title="Obtaining word embedding from existing classification model",
  booktitle="Intelligent Systems Design and Applications",
  year="2018",
  series="ISDA 2017 Intelligent Systems Design and Applications",
  journal="Advances in Intelligent Systems and Computing",
  volume="2018",
  number="736",
  pages="540--547",
  publisher="Springer International Publishing",
  address="Cham",
  doi="10.1007/978-3-319-76348-4\{_}52",
  isbn="978-3-319-76347-7",
  issn="2194-5357",
  url="https://www.fit.vut.cz/research/publication/11546/"
}