Creating and sharing knowledge for telecommunications

Project: Quantum Big Data

Acronym: QbigD
Main Objective:
Dealing with high dimensional and big data is one of the most important problems in many fields of science and engineering.
Addressing this problem has an overwhelming social­economical impact, as the mining procedures are used in a large spectrum of
applications including, clinical diagnosis, network traffic analysis, surveillance and monitoring logs, financial market analysis, web
indexing, among many others.
Following the steps of top research groups and dominant companies in the area, such as Google, the chief objective of this project
is to bring together experts on data mining and quantum information theory to develop new methods for classification and
clustering that can handle big data.
The underlying idea behind using quantum information in data mining is to replace classical probability distributions, thoroughly
used in mining methods, by density operators. To recover a classical probability distribution we need to consider a quantum
observable and apply Born’s rule. At first sight this might seem a naïve approach, however, it was able to justify the performance of
many popular data mining methods whose merits were only granted by empirical evidence. The advantage of using density
operators is twofold: they generalize classical distributions; and they can encode much more complex interference patterns (like
entanglement) that are disregarded classically.
Taking into account the research background and the interests of the interdisciplinary team that composes the project, the key
innovations in the area of data mining proposed in this exploratory project are the following:

1) Learning graphical models endowed with density operators;

2) Classifying with quantum Kolmogorov complexity;

3) Clustering using quantum walk dynamics.
Reference: IT FEDER
Funding: FEDER
Start Date: 01-04-2016
End Date: 01-03-2018
Team: Paulo Alexandre Carreira Mateus, Alexandra Sofia Martins de Carvalho, André Nuno Carvalho Souto, Chrysoula Vlachou, Francisco Miguel Alves Campos de Sousa Dionísio, Nikola Paunkovic
Groups: Pattern and Image Analysis – Lx, Security and Quantum Information - Lx
Partners: IT
Local Coordinator: Paulo Alexandre Carreira Mateus
Associated Publications