Work

Quantum-inspired Probabilistic Database Corruption Detection

Year: 2023

Type: article

Abstract: A crucial problem of machine learning is the management of unclean probabilistic databases. We approach this problem by applying recent results and methods from quantum information to detect a specific class of database corruption. We present a quantifier of the global corruption of the probabilistic database and show its relationship with detection protocols based on generalized Bell inequalities. Furthermore, we show a relation between the noise generating the corruption and information encoded in the database schema. Finally, we discuss how our work indicates a way to export quantum information results to study noise in probabilistic databases.

Source: Parallel Processing Letters

DOI: 10.1142/s0129626423400145

Author: Roberto Salazar

Institution: Jagiellonian University

Language: English

Topic: Quantum Computing Algorithms and Architecture

Subfield: Artificial Intelligence

Field: Computer Science

Domain: Physical Sciences

Open Access status: green

Funder: Fundacja na rzecz Nauki Polskiej

Award: G3116650686