Unimore logo AImageLab
 Gianpaolo Bontempo

Gianpaolo Bontempo

Position at AImageLab:

PhD Student
Dipartimento di Ingegneria "Enzo Ferrari"

Email:

gianpaolo_DOT_bontempo_AT_unimore_DOT_it

Gianpaolo Bontempo

As a dedicated researcher in medical image analysis, my work focuses on two crucial areas: Multi-Instance Learning (MIL) for Whole Slide Image (WSI) classification and Neuro-Symbolic Continual Learning for high-level concept mapping and consistent reasoning.

In the realm of MIL, I have developed a groundbreaking graph-based multi-scale MIL approach known as DAS-MIL. By leveraging self-supervised feature extraction, graph-based architecture, and distillation loss, DAS-MIL overcomes the challenges associated with the pixel-level annotation of gigapixel-sized WSIs. Through extensive experimentation on well-known datasets, my approach has achieved exceptional performance, surpassing state-of-the-art methods.

Furthermore, my research has delved into Neuro-Symbolic Continual Learning, an innovative approach that tackles a sequence of neuro-symbolic tasks by integrating prior knowledge and maintaining reasoning consistency. By effectively leveraging prior knowledge and preserving high-quality concepts over time, COOL has proven its worth in sustained high performance on neuro-symbolic continual learning tasks where other approaches fall short. 


Research Projects


Research Activities


Publications

1 Bontempo, Gianpaolo; Bartolini, Nicola; Lovino, Marta; Bolelli, Federico; Anni, Virtanen; Ficarra, Elisa "Enhancing PFI Prediction with GDS-MIL: A Graph-based Dual Stream MIL Approach" ICIAP 2023: Image Analysis and Processing, Udine, Italy, pp. 1 -12 , Sep 11-15, 2023 Conference
2 Marconato, Emanuele; Bontempo, Gianpaolo; Ficarra, Elisa; Calderara, Simone; Passerini, Andrea; Teso, Stefano "Neuro Symbolic Continual Learning: Knowledge, Reasoning Shortcuts and Concept Rehearsal" 2023 | DOI: 10.48550/arxiv.2302.01242 Other
3 Bontempo, Gianpaolo; Porrello, Angelo; Bolelli, Federico; Calderara, Simone; Ficarra, Elisa "DAS-MIL: Distilling Across Scales for MILClassification of Histological WSIs" Medical Image Computing and Computer Assisted Intervention – MICCAI 2023, Vancouver, 2023-10-08, 2023 Conference
4 Bontempo, Gianpaolo; Lumetti, Luca; Porrello, Angelo; Bolelli, Federico; Calderara, Simone; Ficarra, Elisa "Buffer-MIL: Robust Multi-instance Learning with a Buffer-based Approach" Image Analysis and Processing – ICIAP 2023, Udine, Italy, pp. 1 -12 , Sep 11-15, 2023 Conference
5 Citarrella, Francesca; Bontempo, Gianpaolo; Lovino, Marta; Ficarra, Elisa "FusionFlow: an integrated system workflow for gene fusion detection in genomic samples" European Conference on Advances in Databases and Information Systems, Torino, 2022, 2022 Conference
6 Citarrella, F.; Bontempo, G.; Lovino, M.; Ficarra, E. "FusionFlow: An Integrated System Workflow for Gene Fusion Detection in Genomic Samples" Communications in Computer and Information Science, vol. 1652 CCIS, pp. 79 -88 , 2022 | DOI: 10.1007/978-3-031-15743-1_8 Chapter in Book
7 Marconato, E.; Bontempo, G.; Teso, S.; Ficarra, E.; Calderara, S.; Passerini, A. "Catastrophic Forgetting in Continual Concept Bottleneck Models" Lecture Notes in Computer Science, vol. 13374 LNCS, 23 May 2022through 27 May 2022Code, pp. 539 -547 , lecce, 2022 | DOI: 10.1007/978-3-031-13324-4_46 Conference
8 Lovino, Marta; Bontempo, Gianpaolo; Cirrincione, Giansalvo; Ficarra, Elisa "Multi-omics Classification on Kidney Samples Exploiting Uncertainty-Aware Models" Intelligent Computing Theories and Application, vol. 12464 LNCS, Bari (Online), pp. 32 -42 , October,2020, 2020 | DOI: 10.1007/978-3-030-60802-6_4 Conference