Luca Savant Aira

Ph.D. in Ingegneria Elettrica, Elettronica E Delle Comunicazioni , 38^th cycle (2022-2025)

Ph.D. obtained in 2026

Dissertation:

Deep Learning for 3D World Representations (Abstract)

Tutors:

Enrico Magli Diego Valsesia Giulia Fracastoro

Profile

Research topic

Multi-image restoration exploiting neural implicit world models

Keywords

Multimedia Signal Processing

Big Data, Machine Learning, Neural Networks and Data Science

Biography

I am currently a Ph.D. candidate at the IPL group at Politecnico di Torino, specializing in multi-image restoration by exploiting neural implicit world models. This research allows me to delve into cutting-edge techniques and contribute to the advancement of knowledge in this field.
Prior to my Ph.D., I earned a Master's Degree in Mathematical Engineering from Politecnico di Torino in 2021/2022, with a focus on Statistics and Optimization of Networks. My master's thesis, "Job Scheduling on uniform machines: approximation ratios by computer-assisted proofs," showcased my ability to employ innovative approaches, including computer-assisted proof methods, to achieve accurate and reliable results.
I hold a Bachelor's Degree in Mathematics for Engineering, earned in the period 2017/2020, also from Politecnico di Torino, where I graduated with a thesis titled "Machine Learning Performance Analysis for Discrete Fracture Networks," earning a grade of 110L/110.
My academic journey has instilled in me a profound passion for logic and scientific rigor. I am particularly interested in Machine Learning and Deep Learning, with a focus on neural architectures. Additionally, my expertise extends to Geometry, including Linear Algebra, Topology, and Differential Geometry, as well as Optimization through Mathematical Programming. This multidisciplinary background positions me as a versatile professional ready to contribute to advancements at the intersection of mathematics and technology.

Teaching

Collegi of the degree programmes

Collegio di Ingegneria Informatica, del Cinema e Meccatronica. Componente invitato

Teachings

Master of Science

Signal, image and video processing and learning (modulo di Image and video processing and learning). A.A. 2023/24, COMMUNICATIONS ENGINEERING. Collaboratore del corso
Statistical learning and neural networks. A.A. 2023/24, ICT FOR SMART SOCIETIES (ICT PER LA SOCIETA' DEL FUTURO). Collaboratore del corso

MostraNascondi A.A. passati

Bachelor of Science

Elaborazione di immagine e video. A.A. 2025/26, INGEGNERIA DEL CINEMA E DEI MEZZI DI COMUNICAZIONE. Collaboratore del corso

Publications

Works published during the Ph.D. View all publications in Porto@Iris

Savant Aira, Luca (2026)
Deep Learning for 3D World Representations. relatore: MAGLI, ENRICO; VALSESIA, DIEGO; , 38. XXXVIII Ciclo, P.: 137
Doctoral Thesis
SAVANT AIRA, Luca; Valsesia, Diego; Magli, Enrico (2025)
Modeling Uncertainty for Gaussian Splatting. In: IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS, pp. 1-7. ISSN 2162-237X
Contributo su Rivista
Aira, Luca Savant; Facciolo, Gabriele; Ehret, Thibaud (2025)
Gaussian Splatting for Efficient Satellite Image Photogrammetry. In: 2025 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Nashville (USA), 10-17 June 2025, pp. 5959-5969. ISBN: 979-8-3315-4364-8
Contributo in Atti di Convegno (Proceeding)
Montanaro, A.; SAVANT AIRA, Luca; Aiello, E.; Valsesia, D.; Magli, E. (2024)
MOTIONCRAFT: Physics-based Zero-Shot Video Generation. In: 38th Conference on Neural Information Processing Systems, NeurIPS 2024, Vancouver (Can), 10 - 15 December 2024. ISSN 1049-5258
Contributo in Atti di Convegno (Proceeding)
Aira, Luca Savant; Valsesia, Diego; Molini, Andrea Bordone; Fracastoro, Giulia; Magli, ... (2024)
Deep 3D World Models for Multi-Image Super-Resolution Beyond Optical Flow. In: IEEE ACCESS, vol. 12, pp. 188902-188913. ISSN 2169-3536
Contributo su Rivista