« Raisonnement abstrait » : différence entre les versions
(Page créée avec « == en construction == == Définition == xxxxx Voir aussi '''Corpus d'Abstraction et de Raisonnement''' == Français == ''' XXXXXX''' == Anglais == ''' Abstract Reasoning''' ''' Fluid Intelligence''' ==Sources== [https://arxiv.org/abs/2405.01535 Source : arxiv] Catégorie:vocabulary ») |
Aucun résumé des modifications |
||
Ligne 14 : | Ligne 14 : | ||
''' Fluid Intelligence''' | ''' Fluid Intelligence''' | ||
''Abstract reasoning is the ability to solve complex problems by identifying regularities and relations in the problem being solved and utilizing them for deducing the solution. Changes in how problems are phrased that seem minor to humans and do not affect their performance can render them unsolvable to networks. In a field dominated by extensive training with an extensive number of examples, abstract reasoning models aim to close the gap between human fluid intelligence and the performance of deep neural networks, which typically require extensive prior training. In fact, abstract reasoning is possible even in naïve and random ANNs and these networks can achieve what looks like symbolic abstract reasoning without any training and, hence, any memory recall.'' | |||
==Sources== | ==Sources== |
Version du 16 octobre 2024 à 13:28
en construction
Définition
xxxxx
Voir aussi Corpus d'Abstraction et de Raisonnement
Français
XXXXXX
Anglais
Abstract Reasoning
Fluid Intelligence
Abstract reasoning is the ability to solve complex problems by identifying regularities and relations in the problem being solved and utilizing them for deducing the solution. Changes in how problems are phrased that seem minor to humans and do not affect their performance can render them unsolvable to networks. In a field dominated by extensive training with an extensive number of examples, abstract reasoning models aim to close the gap between human fluid intelligence and the performance of deep neural networks, which typically require extensive prior training. In fact, abstract reasoning is possible even in naïve and random ANNs and these networks can achieve what looks like symbolic abstract reasoning without any training and, hence, any memory recall.