Portrait of Hiba Bensalem

Hiba Bensalem

Position
PhD student, since 2024
Country of origin
Tunisia
Expertise
Spatial statistics, Deep learning for microscopy images, Spatial point processes
Research interest
Advancing AI models for analyzing spatial distributions and patterns of objects in large-scale biological microscopy image data, combining the strengths of spatial statistics and deep learning.
See CV

Hiba Bensalem joined the MOSAIC group in December 2024 as a Ph.D. student. Originally from Tunisia, Hiba built a strong foundation in mathematics and the sciences during her pre-engineering studies, where her high national ranking earned her admission into Tunisia’s top-ranked information and communication technology engineering school, the Higher School of Communication of Tunis (SUP’COM). During her academic journey at SUP’COM, her interest in data-driven modeling and artificial intelligence developed, particularly for image-based applications.

For her Master’s thesis project, Hiba was awarded a scholarship through the IMPRS-CellDevoSys program of the Max Planck Society. For her graduation project, she worked in the MOSAIC group under the supervision of Prof. Sbalzarini for six months. Together with Dominik Sturm from the MOSAIC Group, she developed a spatial point process-informed variational autoencoder. This constitutes a novel representation-learning technique designed to distinguish between different spatial arrangements of objects with respect to each other or to a reference structure. This project marked the beginning of Hiba’s curiosity about integrating spatial statistics with neural networks to tackle challenges in biomedical data science.

Hiba’s PhD project focuses on advancing AI models for analyzing spatial distributions and patterns of objects in large-scale biological microscopy image data, combining the strengths of spatial statistics and deep learning.

Publications

  1. Spatially Informed Autoencoders for Interpretable Visual Representation Learning

    D. Sturm, H. Bensalem and I. F. Sbalzarini

    International Conference on Learning Representations (ICLR), 2026