Sammanfattning
has been studied in computational agent societies with language games.
In this paper we introduce methods that can be used to ground
a language in the needs of the agents, i.e. practicality.
The needs of an agent arise from its goals and environment, which
together dictate what kind of a language is practical.
The methods are tested in multi-agent simulations.
Our results show that with our methods a practical language can emerge.
Interestingly we also observed, that a language does not need to be coherent
in the whole society for it to have practical benefits for the agents.
| Originalspråk | engelska |
|---|---|
| Titel på gästpublikation | AISB Language Learning for Artificial Agents symposium |
| Antal sidor | 5 |
| Utgivningsdatum | apr 2019 |
| Sidor | 16-20 |
| Status | Publicerad - apr 2019 |
| MoE-publikationstyp | A4 Artikel i en konferenspublikation |
| Evenemang | 2019 AISB Convention - Falmouth, Storbritannien Varaktighet: 16 apr 2019 → 18 apr 2019 |
Vetenskapsgrenar
- 113 Data- och informationsvetenskap
Citera det här
}
On Grounding Language Games in Practicality. / Hantula, Otto Albert.
AISB Language Learning for Artificial Agents symposium. 2019. s. 16-20.Forskningsoutput: Kapitel i bok/rapport/konferenshandling › Konferensbidrag › Vetenskaplig › Peer review
TY - GEN
T1 - On Grounding Language Games in Practicality
AU - Hantula, Otto Albert
PY - 2019/4
Y1 - 2019/4
N2 - Emergence of language grounded in perceptionhas been studied in computational agent societies with language games.In this paper we introduce methods that can be used to grounda language in the needs of the agents, i.e. practicality. The needs of an agent arise from its goals and environment, whichtogether dictate what kind of a language is practical.The methods are tested in multi-agent simulations.Our results show that with our methods a practical language can emerge. Interestingly we also observed, that a language does not need to be coherentin the whole society for it to have practical benefits for the agents.
AB - Emergence of language grounded in perceptionhas been studied in computational agent societies with language games.In this paper we introduce methods that can be used to grounda language in the needs of the agents, i.e. practicality. The needs of an agent arise from its goals and environment, whichtogether dictate what kind of a language is practical.The methods are tested in multi-agent simulations.Our results show that with our methods a practical language can emerge. Interestingly we also observed, that a language does not need to be coherentin the whole society for it to have practical benefits for the agents.
KW - 113 Computer and information sciences
KW - multi-agent systems
KW - language game
M3 - Conference contribution
SP - 16
EP - 20
BT - AISB Language Learning for Artificial Agents symposium
ER -