Lugus 1214685 - Call for Research Grant Assignment - PS1660

Descrizione

Lugus 1214685 - Call for Research Grant Assignment - PS1660 Sintesi in italiano (traduzione automatica): L'Università di Coimbra ha aperto una call per l'assegnazione di tre borse di ricerca post-dottorato nel progetto PS1660, intitolato 'Plataforma de bilhética'. I candidati devono possedere un dottorato in Ingegneria Informatica conseguito negli ultimi tre anni e non devono aver svolto la maggior parte della loro ricerca presso l'istituzione ospitante. La posizione è situata nel Dipartimento di Ingegneria Informatica della Facoltà di Scienze e Tecnologia. Il progetto si concentra sull'uso delle Reti Neurali Group-Equivariant per l'apprendimento di rappresentazioni robuste in scenari di fusione di dati multimodali. La durata iniziale della borsa è di 3 mesi, con un compenso mensile di € 1901,00. Call for Postdoctoral Research Assignment The University of Coimbra opens a call for the assignment of a postdoctoral research grant, with three position(s), under the project PS1660, title “Plataforma de bilhética”. The call is governed by this Notice of Opening, the University of Coimbra Research Grant Regulations (RBI-UC), subsidiarily by the Research Grant Regulations of the Foundation for Science and Technology (RBI-FCT), by the Research Fellow Statute (EBI), by the Norms for Grant Assignment and Management within R&D Projects, by the Code of Administrative Procedure (CPA), all in their current wording, and by other applicable national and community legislation. Procedure Reference: IT137-26-304 I - Legal admission requirements I.I - Grant Recipients: Holders of a doctoral degree in the scientific area of Informatics Engineering, obtained within the three years prior to the start date of the grant, where the research work leading to the degree was carried out at a host institution different from the one where the work of the current grant will be conducted. I.II – Ineligible candidates: The following candidates will not be admitted to the call:  Candidates who, with the award of the current grant, including possible renewals, exceed a cumulative period of three years in this type of grant, whether consecutive or intermittent.  Candidates who obtained their doctoral degree more than three years before the application submission date.  Candidates who have conducted the majority of the research work leading to their doctoral degree at the host institution of the current grant. I.III - Eligible for grant assignment if meeting the requirements: a) Portuguese nationals or nationals of other EU Member States b) Nationals of third countries c) Stateless persons d) Beneficiaries of political refugee status II - Grant location: Department of Computer Engineering, Faculty of Sciences and Technology, University of Coimbra III- Scientific supervision/coordination of the grant: Pedro Manuel Henriques da Cunha Abreu IV - Work Plan / Goals to be achieved: Title: Group-Equivariant Feature Learning for Multi-Modal Data Fusion This project aims to investigate the use of Group-Equivariant Neural Networks (G-Equivariant NNs) for learning robust feature representations in multi-modal data fusion scenarios. Work Plan: Month 1 – Literature Review and Experimental Framework -Conduct an in-depth review of recent advances in Group-Equivariant Neural Networks and geometric deep learning. -Review state-of-the-art multi-modal fusion architectures and identify representative benchmark datasets. -Select appropriate equivariant architectures (e.g., E(2)-CNNs, Steerable CNNs, Lie Group CNNs) and define the experimental protocol. -Implement the experimental environment and reproduce selected baseline methods. Month 2 – Development of the Equivariant Fusion Model -Design and implement an equivariant feature extraction module for multiple sensing modalities. -Investigate different feature fusion strategies (early, intermediate, and late fusion). -Train and optimize the proposed architecture using selected datasets. -Evaluate robustness under geometric transformations and limited-data scenarios. Month 3 – Validation, Analysis and Dissemination -Perform a comprehensive comparison against conventional CNN-based fusion methods. -Analyze computational complexity, robustness, and generalization capabilities. -Prepare a scientific manuscript describing the proposed methodology and experimental results. -Release reproducible code and documentation. V - Initial grant duration: 3 months V.I - Renewal Possibility: Non-renewable VI - Funding and financial conditions of the grant VI.I - Monthly grant amount (paid by bank transfer at the end of each month): € 1901,00 VI.II - In addition to the above amounts, voluntary social security (SSV) is included when the grant has a duration of six months or more, corresponding to the first level, if the candidate opts for it, as Annuncio in inglese. Fonte: Euraxess (Commissione europea).