A pre-doctoral contract is offered to train a doctoral candidate under the FI-SDUR call from AGAUR. This position is available at the Graphics and Imaging Laboratory (GILAB) of the University of Girona. The thesis will be carried out under the supervision of a senior researcher in the group.
Research Area: This PhD project analyses deep neural networks (DNNs) using information theory to uncover their performance and operation principles.
Objectives:
- Investigate DNNs through the lens of information theory, considering networks as Markov models.
- Explore the information bottleneck hypothesis in convolutional neural networks (CNNs) to understand their learning process.
- Apply information theory tools to image analysis, including fusion of unimodal and multimodal images, and color-to-gray transformations.
- Develop and test new information-theoretic measures as cost functions and for dimensionality reduction in deep learning.
Expected Outcomes:
- Deeper insights into the internal workings of DNNs.
- Enhanced methods for image segmentation, clustering, and image retrieval using information theory.
- Improved data processing and feature extraction efficiency within various applications, including visual computing and economics.
Required Education Level
Applicants must have completed the studies granting access to official doctoral programs between January 2021, and July 2024, or at the time of application, applicants must be enrolled in a degree program that allows them to enrol in the doctoral program for the 2023-2024 academic year.
- Computer Science: Master’s Degree or equivalent with a minimum average grade of 6.5 (on a scale of 0 to 10)
- Mathematics: Master’s Degree or equivalent with a minimum average grade of 6.5 (on a scale of 0 to 10)
- For studies completed abroad, the average grade of the academic transcript must be calculated according to the current resolutions of the Ministry of Universities for grade equivalencies.
Skills/Qualifications
- Experience and knowledge in neural networks with TensorFlow, PyTorch or other libraries
- Previous experience with programming (e.g., Matlab, C++, Python, etc.) is highly valued
- Experience and knowledge in image processing is a plus
- A good level of English is an asset (minimum B2)
Contract Conditions: The conditions will align with the FI-SDUR call from AGAUR. The main conditions are:
- Duration: 3 years
- Remuneration: €1,386.57/month gross for the 1st and 2nd year, and €1,485.61/month gross for the 3rd year
- Grant: €3,000 for training and internationalization activities
- Start Date: The contract must start from 1 April 2025. Exceptionally, non-EU nationals may join until 1 October 2025.
How to Apply >> Send your CV, academic record*, and motivation letter to cristina.diaz@udg.edu by 14 July 2023. We will support selected candidates in applying for the FI-SDUR fellowship at the AGAUR site.
* For studies completed abroad, the average grade of the academic transcript must be calculated according to the current resolutions of the Ministry of Universities for grade equivalencies.
Please ensure that you meet all requirements for pre-enrolment for the doctoral program at the Universitat de Girona (here).


Adrià Julià, researcher at the GILAB research group and member of the Institute of Informatics and Applications, obtained a PhD on 11 November. His dissertation was supervised by Imma Boada and Marc Ruiz and is entitled “Medical imaging techniques for the analysis and measurement of composite materials”.
Núria Banús, researcher at the GILAB research group and member of the Institute of Informatics and Applications, obtained a PhD on 16 December. Her dissertation was supervised by Imma Boada and Pau Xiberta and is entitled “New solutions to control robotic environments: Quality control in food packaging”.
the rehabilitation of stroke patients. The project is led by researcher Yolanda Silva from the Hospital Dr. Josep Trueta, with the collaboration of Imma Boada, of the Graphics and Imaging Laboratory (GILAB) and Eva Barris, of MUTUAM Girona.


Pau Xiberta, researcher at the GILAB research group and member of the Institute of Informatics and Applications, obtained a PhD on 25 May. His dissertation was supervised by Anton Bardera and Imma Boada and is entitled ‘Medical imaging applied to teaching and meat science’.