Semi-Supervised Learning using Differentiable Reasoning

Research output: Contribution to JournalSpecial issue (editing)Academicpeer-review

Abstract

We introduce Differentiable Reasoning (DR), a novel semi-supervised learning technique which uses relational background knowledge to benefit from unlabeled data. We apply it to the Semantic Image Interpretation (SII) task and show that background knowledge provides significant improvement. We find that there is a strong but interesting imbalance between the contributions of updates from Modus Ponens (MP) and its logical equivalent Modus Tollens (MT) to the learning process, suggesting that our approach is very sensitive to a phenomenon called the Raven Paradox. We propose a solution to overcome this situation.
Original languageEnglish
Pages (from-to)633-653
Number of pages20
JournalIFCoLog Journal of Logic and its Applications
Volume6
Issue number4
Publication statusAccepted/In press - 2019

Fingerprint

Supervised learning
Semantics

Cite this

@article{aa21302d28c4436aac754a7ac90bfacf,
title = "Semi-Supervised Learning using Differentiable Reasoning",
abstract = "We introduce Differentiable Reasoning (DR), a novel semi-supervised learning technique which uses relational background knowledge to benefit from unlabeled data. We apply it to the Semantic Image Interpretation (SII) task and show that background knowledge provides significant improvement. We find that there is a strong but interesting imbalance between the contributions of updates from Modus Ponens (MP) and its logical equivalent Modus Tollens (MT) to the learning process, suggesting that our approach is very sensitive to a phenomenon called the Raven Paradox. We propose a solution to overcome this situation.",
author = "{van Krieken}, Emile and E. Acar and {van Harmelen}, Frank",
year = "2019",
language = "English",
volume = "6",
pages = "633--653",
journal = "IFCoLog Journal of Logic and its Applications",
number = "4",

}

Semi-Supervised Learning using Differentiable Reasoning. / van Krieken, Emile; Acar, E.; van Harmelen, Frank.

In: IFCoLog Journal of Logic and its Applications, Vol. 6, No. 4, 2019, p. 633-653.

Research output: Contribution to JournalSpecial issue (editing)Academicpeer-review

TY - JOUR

T1 - Semi-Supervised Learning using Differentiable Reasoning

AU - van Krieken, Emile

AU - Acar, E.

AU - van Harmelen, Frank

PY - 2019

Y1 - 2019

N2 - We introduce Differentiable Reasoning (DR), a novel semi-supervised learning technique which uses relational background knowledge to benefit from unlabeled data. We apply it to the Semantic Image Interpretation (SII) task and show that background knowledge provides significant improvement. We find that there is a strong but interesting imbalance between the contributions of updates from Modus Ponens (MP) and its logical equivalent Modus Tollens (MT) to the learning process, suggesting that our approach is very sensitive to a phenomenon called the Raven Paradox. We propose a solution to overcome this situation.

AB - We introduce Differentiable Reasoning (DR), a novel semi-supervised learning technique which uses relational background knowledge to benefit from unlabeled data. We apply it to the Semantic Image Interpretation (SII) task and show that background knowledge provides significant improvement. We find that there is a strong but interesting imbalance between the contributions of updates from Modus Ponens (MP) and its logical equivalent Modus Tollens (MT) to the learning process, suggesting that our approach is very sensitive to a phenomenon called the Raven Paradox. We propose a solution to overcome this situation.

UR - https://drive.google.com/file/d/1lh3zIsO0cAnlNUv6FNivE_JJKHwlTyhq/view

M3 - Special issue (editing)

VL - 6

SP - 633

EP - 653

JO - IFCoLog Journal of Logic and its Applications

JF - IFCoLog Journal of Logic and its Applications

IS - 4

ER -