HAL will be down for maintenance from Friday, June 10 at 4pm through Monday, June 13 at 9am. More information
Skip to Main content Skip to Navigation

Meaningful objective frequency-based interesting pattern mining

Thomas Delacroix 1, 2
Lab-STICC - Laboratoire des sciences et techniques de l'information, de la communication et de la connaissance : UMR6285
Abstract : In this thesis, we study objective interesting pattern mining processes on datasets such as used in itemset mining. We focus on the notions of objectivity and meaningfulness in mining processes. We establish a link between the meaningfulness of a mining process and that of its corresponding mathematical modeling. We formulate a number of recommendations in terms of modeling choices for increasing both meaningfulness and objectivity. We also establish a link between the study of objective interesting pattern mining and the issue of the automation of scientific discovery.Our theoretical analysis exhibits the adequacy of considering maximum entropy models in such mining processes. We then proceed with presenting a novel mathematical construction for such models, based on a notion of constrained independence, which is specifically adapted to the itemset context. Based on this construction and on tools from algebraic geometry, we present an exact method for computing maximum entropy models. Last, based on our recommendations for the mathematical modeling of pattern mining processes and our notion of constrained independence, we present a new objective interesting pattern mining algorithm.
Document type :
Complete list of metadata

Contributor : Abes Star :  Contact
Submitted on : Thursday, July 15, 2021 - 8:09:09 AM
Last modification on : Monday, April 4, 2022 - 9:28:31 AM
Long-term archiving on: : Saturday, October 16, 2021 - 6:08:57 PM


Version validated by the jury (STAR)


  • HAL Id : tel-03286641, version 1


Thomas Delacroix. Meaningful objective frequency-based interesting pattern mining. Artificial Intelligence [cs.AI]. Ecole nationale supérieure Mines-Télécom Atlantique, 2021. English. ⟨NNT : 2021IMTA0251⟩. ⟨tel-03286641⟩



Record views


Files downloads