Towards reasoning based representations

Deep Consistence Seeking Machine

A. Lőrincz, M. Csákvári, Fóthi, Z. Milacski, A. Sárkány, Z. Tősér

Research output: Contribution to journalArticle

Abstract

Machine learning is making substantial progress in diverse applications. The success is mostly due to advances in deep learning. However, deep learning can make mistakes and its generalization abilities to new tasks are questionable. We ask when and how one can combine network outputs, when (i) details of the observations are evaluated by learned deep components and (ii) facts and rules are available. The Deep Consistence Seeking (DCS) machine seeks for consistent and deterministic event descriptions and improves the representation accordingly. The machine has an anomaly detection component that may trigger coherence seeking. Coherence seeking resolves conflicts between computational modules by preferring components with higher scores. We illustrate that context can help in correcting recognitions and in deriving training samples for self-training. We put these concepts into a general framework of cognition, by distinguishing creativity, rule extraction, verification, and symbol grounding. We demonstrate our approach in a driving scenario.

Original languageEnglish
Pages (from-to)92-108
Number of pages17
JournalCognitive Systems Research
Volume47
DOIs
Publication statusPublished - Jan 1 2018

Fingerprint

Learning
Aptitude
Creativity
Electric grounding
Cognition
Learning systems
Deep learning
Conflict (Psychology)
Recognition (Psychology)
Machine Learning
Generalization (Psychology)

Keywords

  • Communication
  • Complexity
  • Deep learning
  • Episodic description
  • Recognition by components
  • Rule-based system

ASJC Scopus subject areas

  • Experimental and Cognitive Psychology
  • Cognitive Neuroscience
  • Artificial Intelligence

Cite this

Towards reasoning based representations : Deep Consistence Seeking Machine. / Lőrincz, A.; Csákvári, M.; Fóthi; Milacski, Z.; Sárkány, A.; Tősér, Z.

In: Cognitive Systems Research, Vol. 47, 01.01.2018, p. 92-108.

Research output: Contribution to journalArticle

Lőrincz, A. ; Csákvári, M. ; Fóthi ; Milacski, Z. ; Sárkány, A. ; Tősér, Z. / Towards reasoning based representations : Deep Consistence Seeking Machine. In: Cognitive Systems Research. 2018 ; Vol. 47. pp. 92-108.
@article{0aef5b3b4eb54cee8beb2c1a44589505,
title = "Towards reasoning based representations: Deep Consistence Seeking Machine",
abstract = "Machine learning is making substantial progress in diverse applications. The success is mostly due to advances in deep learning. However, deep learning can make mistakes and its generalization abilities to new tasks are questionable. We ask when and how one can combine network outputs, when (i) details of the observations are evaluated by learned deep components and (ii) facts and rules are available. The Deep Consistence Seeking (DCS) machine seeks for consistent and deterministic event descriptions and improves the representation accordingly. The machine has an anomaly detection component that may trigger coherence seeking. Coherence seeking resolves conflicts between computational modules by preferring components with higher scores. We illustrate that context can help in correcting recognitions and in deriving training samples for self-training. We put these concepts into a general framework of cognition, by distinguishing creativity, rule extraction, verification, and symbol grounding. We demonstrate our approach in a driving scenario.",
keywords = "Communication, Complexity, Deep learning, Episodic description, Recognition by components, Rule-based system",
author = "A. Lőrincz and M. Cs{\'a}kv{\'a}ri and F{\'o}thi and Z. Milacski and A. S{\'a}rk{\'a}ny and Z. Tős{\'e}r",
year = "2018",
month = "1",
day = "1",
doi = "10.1016/j.cogsys.2017.08.004",
language = "English",
volume = "47",
pages = "92--108",
journal = "Cognitive Systems Research",
issn = "1389-0417",
publisher = "Elsevier",

}

TY - JOUR

T1 - Towards reasoning based representations

T2 - Deep Consistence Seeking Machine

AU - Lőrincz, A.

AU - Csákvári, M.

AU - Fóthi,

AU - Milacski, Z.

AU - Sárkány, A.

AU - Tősér, Z.

PY - 2018/1/1

Y1 - 2018/1/1

N2 - Machine learning is making substantial progress in diverse applications. The success is mostly due to advances in deep learning. However, deep learning can make mistakes and its generalization abilities to new tasks are questionable. We ask when and how one can combine network outputs, when (i) details of the observations are evaluated by learned deep components and (ii) facts and rules are available. The Deep Consistence Seeking (DCS) machine seeks for consistent and deterministic event descriptions and improves the representation accordingly. The machine has an anomaly detection component that may trigger coherence seeking. Coherence seeking resolves conflicts between computational modules by preferring components with higher scores. We illustrate that context can help in correcting recognitions and in deriving training samples for self-training. We put these concepts into a general framework of cognition, by distinguishing creativity, rule extraction, verification, and symbol grounding. We demonstrate our approach in a driving scenario.

AB - Machine learning is making substantial progress in diverse applications. The success is mostly due to advances in deep learning. However, deep learning can make mistakes and its generalization abilities to new tasks are questionable. We ask when and how one can combine network outputs, when (i) details of the observations are evaluated by learned deep components and (ii) facts and rules are available. The Deep Consistence Seeking (DCS) machine seeks for consistent and deterministic event descriptions and improves the representation accordingly. The machine has an anomaly detection component that may trigger coherence seeking. Coherence seeking resolves conflicts between computational modules by preferring components with higher scores. We illustrate that context can help in correcting recognitions and in deriving training samples for self-training. We put these concepts into a general framework of cognition, by distinguishing creativity, rule extraction, verification, and symbol grounding. We demonstrate our approach in a driving scenario.

KW - Communication

KW - Complexity

KW - Deep learning

KW - Episodic description

KW - Recognition by components

KW - Rule-based system

UR - http://www.scopus.com/inward/record.url?scp=85034101865&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85034101865&partnerID=8YFLogxK

U2 - 10.1016/j.cogsys.2017.08.004

DO - 10.1016/j.cogsys.2017.08.004

M3 - Article

VL - 47

SP - 92

EP - 108

JO - Cognitive Systems Research

JF - Cognitive Systems Research

SN - 1389-0417

ER -