Powered by OpenAIRE graph
Found an issue? Give us feedback
ZENODOarrow_drop_down
ZENODO
Conference object . 2026
License: CC BY
Data sources: Datacite
ZENODO
Article . 2026
License: CC BY
Data sources: Datacite
ZENODO
Article . 2026
License: CC BY
Data sources: Datacite
ZENODO
Conference object . 2026
License: CC BY
Data sources: Datacite
versions View all 4 versions
addClaim

DH on the Edge?

Authors: Brolich, Nina;

DH on the Edge?

Abstract

Mit Tiny ML, Edge AI oder Efficient AI etabliert sich parallel zu LLMs ein Forschungsfeld, das auf die Entwicklung kleiner, energie- und kosteneffizienter und lokal ausführbarer Deep-Learning-Modelle abzielt. Zentral ist hier das Deployment auf Edge Devices (z.B. Mikrocontroller, Raspberrry Pi, Smartwatches etc.). Für die DH ergeben sich Vorteile primär aus der Möglichkeit der dezentralen Datenverarbeitung direkt am Erfassungsort sowie der leichteren und günstigeren Verfügbarkeit von entsprechender Hardware gegenüber HPC-Infrastruktur oder fremdgehosteten, potentiell kostenpflichtigen LLMs, z.B. in archäologischer Feldforschung, partizipativen Citizen-Science- oder Oral-History-Projekten oder im Bereich des kulturellen Erbes. Für diesen Beitrag wurden Mistral, BERT und mehrere komprimierte und quantisierte BERT-Modelle im Kontext eines Named-Entity-Recognition-Tasks miteinander verglichen. Dabei zeigte sich, dass komprimierte Modelle nur marginal an Performance verlieren, aber deutlich kleiner und schneller sind, sodass Edge Deployment vorstellbar ist. Die Performance von Mistral war in diesem Beispiel deutlich schlechter als die Performance der spezialiserten BERT-Modelle.

Keywords

Paper, benannte Entitäten (named entities), Hardware, DHd2026, Efficient AI, Programming, Minimal Computing, Poster, Edge AI, Assessing, Annotating, Theorising

  • BIP!
    Impact byBIP!
    selected citations
    These citations are derived from selected sources.
    This is an alternative to the "Influence" indicator, which also reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically).
    0
    popularity
    This indicator reflects the "current" impact/attention (the "hype") of an article in the research community at large, based on the underlying citation network.
    Average
    influence
    This indicator reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically).
    Average
    impulse
    This indicator reflects the initial momentum of an article directly after its publication, based on the underlying citation network.
    Average
Powered by OpenAIRE graph
Found an issue? Give us feedback
selected citations
These citations are derived from selected sources.
This is an alternative to the "Influence" indicator, which also reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically).
BIP!Citations provided by BIP!
popularity
This indicator reflects the "current" impact/attention (the "hype") of an article in the research community at large, based on the underlying citation network.
BIP!Popularity provided by BIP!
influence
This indicator reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically).
BIP!Influence provided by BIP!
impulse
This indicator reflects the initial momentum of an article directly after its publication, based on the underlying citation network.
BIP!Impulse provided by BIP!
0
Average
Average
Average
Upload OA version
Are you the author of this publication? Upload your Open Access version to Zenodo!
It’s fast and easy, just two clicks!