Research projects

Current projects

Fuzzy segmentation
(Utilization of non-supervised learning techniques for client portfolio segmentation)

Obtaining of non-trivial segmentation using qualitative and fuzzy information to improve the relationship between customers and distributors.

Collaborators: Grifone (Textil Seu, S.A.), Instituto Pond's, Supsa (Supermercats Plus-Fresc)

Project Sensorial

Analyzes the use of Support Vector Machines in tasks involving expert knowledge and recommendation, normally performed by experts with specific sensory and cognitive capacities (i.e. colour or perfume designers, sommeliers, etc.).

In addition to the theoretical interest regarding core functions in non-quantitative spaces and the new support-vector-machine-based architectures under development, the project will have a clear impact on managerial applications. On the one hand, in industrial processes in which decision making is provided by experts with specific sensory capacities On the other hand, product recommendation.

Project funded by the MEC (Ministry of Education and Science, TIN2010-20966-C02-01)


 

Project Valora

Qualitative Reasoning techniques are required in high-level cognitive processes where information is not purely numeric and there is incomplete knowledge involved, such as in evaluation and learning environments. These processes frequently occur in situations where sensorial capacities or refined expert knowledge is needed.

This project aims to model, on the one hand, systems where only an assessor or expert is involved and, on the other hand, systems where the evaluation process is conducted with the assistance of a group of evaluators. The study of collective learning resulting from collaboration of a large number of individual agents will also be considered.

Project funded by the MEC (Ministry of Education and Science,  TIN2009-13902-C02-01).

Concluded Projects

MERITO
(Analysis and development of innovative soft-computing techniques with the integration of expert knowledge: an application for financial credit risk measurement)

Assessment of a company's credit risk, reproducing analysts' reasoning. Analysis based on financial ratios and knowledge of the environment.

Project funded by the MEC (Ministry of Education and Science, TIC2002-04371-C02).

AURA
(Automatic learning systems with qualitative reasoning: intelligent support tools for decision-making, applied to finances and marketing)

Design of soft-computing techniques for support in the decision-making process: applications in marketing and finances. The project focuses on the resolution of problems in which there is qualitative and defined information in non-structured environments. The techniques make it possible not only to incorporate the knowledge of experts, but also to extract knowledge and prognoses.

Project funded by the MEC (Ministry of Education and Science, TIN2005-08873-C02).

Expert knowledge management
(knowledge representation techniques and synthesis of experts' information)

Formalisation, study and assessment of experts' knowledge. In particular, studies have been carried out for: the accreditation of pharmaceutical laboratories in doping controls; colour adjustment in the automobile industry; analysis of civic satisfaction in terms of sustainability; and the assessment of value in construction projects.

Collaborators: IMIM (Institut Municipal d'Investigacions Mèdiques), PPG Ibérica, S.A., Ajuntament de Vilanova i la Geltrú, Departament d'Enginyeria de la Construcció de la UPC.