loading page

KitBit: An AI Cognitive Model for Solving Intelligence Test and Numerical Series
  • José Manuel Gilpérez Aguilar ,
  • Victor Corsino ,
  • Luis Herrera
José Manuel Gilpérez Aguilar
Universidad de Castilla La Mancha - UCLM

Corresponding Author:[email protected]

Author Profile
Victor Corsino
Universidad de Castilla La Mancha - UCLM
Author Profile
Luis Herrera
Author Profile

Abstract

The resolution of intelligence tests and in particular the numerical sequences have been analyzed through differentcognitive models and other approaches. In this article we present a new artificial intelligence computational model called KitBit, whichaims to mimic some of the reasoning process that humans use to solve problems. KitBit uses a reduced set of actions and theircombinations to build a predictive model that finds the underlying pattern of different numerical sequences, such as those included inintelligence tests and other much more complex. The system is capable of solving those problems in less than a second with standardcomputational power. The KitBit algorithms have been applied in the OEIS numerical sequence database, reaching the highest numberof series solved to date. Likewise, it is capable of solving the intelligence tests reported by previous models, according to the literature.In this article we expose the fundamentals of the system and its application in finding the patterns of numerical series in general, andmore specifically in cases where these series are part of intelligent tests, even though KitBit has demonstrated its ability to solve othertypes of Intelligence Quotient (IQ) problems, which will be addressed in later publications.
01 Nov 2023Published in IEEE Transactions on Pattern Analysis and Machine Intelligence volume 45 issue 11 on pages 13893-13903. 10.1109/TPAMI.2023.3298592