Ivan Marisca

Ph.D. Student

Languages

Italian (Native)
English (Professional)
French (Basic)
Spanish (Basic)


About me I am a Ph.D. student at the Graph Machine Learning Group, within the Swiss AI lab IDSIA and USI - Università della Svizzera italiana (Switzerland), under the supervision of Prof. Cesare Alippi. Additionally, I am enrolled in the ELLIS Ph.D. program under the joint supervision of Prof. Cesare Alippi and Prof. Michael Bronstein. I am currently on a research visit at the University of Oxford, collaborating with Prof. Michael Bronstein. Previously, I obtained BSc (2017) and MSc (2020) degrees in Computer Science and Engineering at Politecnico di Milano (IT). My master thesis project has been supervised by Prof. Nicola Gatti. My research focuses on graph deep learning for irregular spatiotemporal data. I study the problems of imputation, regularization, and prediction of data coming over time from both physical and virtual sensor networks. Education Ph.D. Student in Informatics 2020 — ongoing Università della Svizzera italiana (USI) Currently, I am a Ph.D. Student at the Swiss AI Lab IDSIA at USI Università della Svizzera Italiana, under the supervision of Prof. Cesare Alippi. Research visit Mar 2024 — Aug 2024 University of Oxford 6-month research visit within the group of Prof. Michael Bronstein. MSc in Computer Science and Engineering 2017 — 2020 Politecnico di Milano Master’s degree with honors (110/110L), defending a thesis on machine learning. During the two years of studies, I mostly attended AI-oriented courses. Exchange Sep 2018 — Jan 2019 Universitat Politècnica de València During the semester spent abroad – in Valencia – within the Erasmus program, I attended Spanish and English courses on programming, robotics and AI. BSc in Engineering of Computing Systems 2014 — 2017 Politecnico di Milano The course program covered general topics of engineering and computer science. High School in Mathematics 2009 — 2014 Liceo C. Caminiti High school diploma with a specific focus in mathematics and science at Liceo Scientifico Caminiti in Santa Teresa di Riva (IT). Academic activities Teaching Advanced Topics in Machine Learning (TA) – MSc at USI Sep 2023 — Jan 2024 MSc at USI Teaching assistant, involved in course organization, lecture preparation and student tutoring. Graph Deep Learning (TA) – MSc at USI Feb 2023 — Jun 2023 MSc at USI Lectures design and students tutoring on team projects. Advanced Topics in Machine Learning (TA) – MSc at USI Sep 2022 — Jan 2023 MSc at USI Students tutoring for projects on reproducibility. Graph Deep Learning (TA) – MSc at USI Feb 2022 — Jun 2022 MSc at USI I gave a lecture on Spatiotemporal Graph Neural Networks and tutored students on projects. Introduzione all’Intelligenza Artificiale e ML (TA) – MSc at USI Sep 2021 — Jan 2022 MSc at USI Course on AI and ML delivered in Italian for high school teachers. Machine Learning (TA) – BSc at USI Feb 2021 — Jun 2021 BSc at USI Lab sessions on practical aspects and show how to design machine learning solutions to real-world problems. Supervised students Marco Latella, MSc at USI 2022 Graph Representation Learning for Multi-site Photovoltaic Energy Production Talks Reading group presentation at TGL reading group (Virtual) Feb 2024 Invited to present the paper Taming Local Effects in Graph-based Spatiotemporal Forecasting at the Temporal Graph Learning Reading Group. (📹 Video) Tutorial at ECML/PKDD (Turin) Sep 2023 Presentation of the tutorial Graph Deep Learning for Spatiotemporal Time Series with colleagues Daniele Zambon and Andrea Cini. Reading group presentation at TGL reading group (Virtual) Apr 2023 Invited to present the paper Scalable Spatiotemporal Graph Neural Networks at the Temporal Graph Learning Reading Group. Spotlight presentation at TGL Workshop (New Orleans) Dec 2022 The Temporal Graph Learning Workshop at NeurIPS 2022. Poster presentation at NeurIPS (New Orleans) Nov 2022 The 36th Conference on Neural Information Processing Systems. Invited talk at Baker Hughes (Virtual) Jul 2022 Invited to give a webinar on time series imputation with colleague Andrea Cini. Poster presentation at ICLR (Virtual) Apr 2022 The 10th International Conference on Learning Representations. Abstract presentation at MLDM (Virtual) Nov 2021 The 10th Italian Workshop on Machine Learning and Data Mining. Awards & Scholarships Doctoral Mobility — Università della Svizzera italiana Dec 2023 CHF 20’000 (≈$23K) to cover the expenses for a 6-month research stay at University of Oxford. Travel Award — NeurIPS Dec 2023 Travel award to attend the NeurIPS conference in New Orleans (US). Best Paper Award — Temporal Graph Learning Workshop @ NeurIPS Dec 2022 For the paper "Scalable Spatiotemporal Graph Neural Networks". Travel Award — NeurIPS Nov 2022 Travel award to attend the NeurIPS conference in New Orleans (US). Scholarship — National Association SAPAR 2019 Scholarship awarded to the top-4 students in STEM subjects. Scholarship — Politecnico di Milano 2019 Reduced tuition for high merits. Projects I believe in worldwide accessibility of science. As such, I make the software I develop for my research publicly available through my GitHub page. You can also find the code related to my publications on the GitHub page of Graph Machine Learning Group. Torch Spatiotemporal Torch Spatiotemporal (TSL) is a library built upon PyTorch and PyG for neural spatiotemporal data processing, with a focus on Graph Neural Networks. GitHub Documentation Other projects GraPV Sep 2020 — Feb 2023 Developing of graph-based methods for multi-site photovoltaic power forecasting, to improve accuracy on portfolio production prediction. The solution is based on novel graph-based AI strategies exploiting existing heterogeneous information and related dependencies. Joint project in collaboration with DXT Commodities, funded by Innosuisse. Publications Graph-based Forecasting with Missing Data through Spatiotemporal Downsampling Ivan Marisca , Cesare Alippi , Filippo Maria Bianchi Preprint, 2024 Graph Deep Learning for Time Series Forecasting Andrea Cini , Ivan Marisca , Daniele Zambon , Cesare Alippi Preprint, 2023 Taming Local Effects in Graph-based Spatiotemporal Forecasting Andrea Cini ^* , Ivan Marisca ^* , Daniele Zambon , Cesare Alippi Advances in Neural Information Processing Systems, 2023 Scalable Spatiotemporal Graph Neural Networks Andrea Cini ^* , Ivan Marisca ^* , Filippo Maria Bianchi , Cesare Alippi Proceedings of the AAAI conference on artificial intelligence, 2023 Learning to Reconstruct Missing Data from Spatiotemporal Graphs with Sparse Observations Ivan Marisca ^* , Andrea Cini ^* , Cesare Alippi Advances in Neural Information Processing Systems, 2022 Filling the G_ap_s: Multivariate Time Series Imputation by Graph Neural Networks Andrea Cini ^* , Ivan Marisca ^* , Cesare Alippi International Conference on Learning Representations, 2022 ^*Equal contribution.

About me

I am a Ph.D. student at the Graph Machine Learning Group, within the Swiss AI lab IDSIA and USI - Università della Svizzera italiana (Switzerland), under the supervision of Prof. Cesare Alippi. Additionally, I am enrolled in the ELLIS Ph.D. program under the joint supervision of Prof. Cesare Alippi and Prof. Michael Bronstein.

I am currently on a research visit at the University of Oxford, collaborating with Prof. Michael Bronstein.

Previously, I obtained BSc (2017) and MSc (2020) degrees in Computer Science and Engineering at Politecnico di Milano (IT). My master thesis project has been supervised by Prof. Nicola Gatti.

My research focuses on graph deep learning for irregular spatiotemporal data. I study the problems of imputation, regularization, and prediction of data coming over time from both physical and virtual sensor networks.

Education

Ph.D. Student in Informatics

2020 — ongoing

Università della Svizzera italiana (USI)

Currently, I am a Ph.D. Student at the Swiss AI Lab IDSIA at USI Università della Svizzera Italiana, under the supervision of Prof. Cesare Alippi.

Research visit

Mar 2024 — Aug 2024

University of Oxford

6-month research visit within the group of Prof. Michael Bronstein.

MSc in Computer Science and Engineering

2017 — 2020

Politecnico di Milano

Master’s degree with honors (110/110L), defending a thesis on machine learning. During the two years of studies, I mostly attended AI-oriented courses.

Exchange

Sep 2018 — Jan 2019

Universitat Politècnica de València

During the semester spent abroad – in Valencia – within the Erasmus program, I attended Spanish and English courses on programming, robotics and AI.

BSc in Engineering of Computing Systems

2014 — 2017

Politecnico di Milano

The course program covered general topics of engineering and computer science.

High School in Mathematics

2009 — 2014

Liceo C. Caminiti

High school diploma with a specific focus in mathematics and science at Liceo Scientifico Caminiti in Santa Teresa di Riva (IT).

Academic activities

Teaching

Advanced Topics in Machine Learning (TA) – MSc at USI

Sep 2023 — Jan 2024

MSc at USI

Teaching assistant, involved in course organization, lecture preparation and student tutoring.
Graph Deep Learning (TA) – MSc at USI

Feb 2023 — Jun 2023

MSc at USI

Lectures design and students tutoring on team projects.
Advanced Topics in Machine Learning (TA) – MSc at USI

Sep 2022 — Jan 2023

MSc at USI

Students tutoring for projects on reproducibility.
Graph Deep Learning (TA) – MSc at USI

Feb 2022 — Jun 2022

MSc at USI

I gave a lecture on Spatiotemporal Graph Neural Networks and tutored students on projects.
Introduzione all’Intelligenza Artificiale e ML (TA) – MSc at USI

Sep 2021 — Jan 2022

MSc at USI

Course on AI and ML delivered in Italian for high school teachers.
Machine Learning (TA) – BSc at USI

Feb 2021 — Jun 2021

BSc at USI

Lab sessions on practical aspects and show how to design machine learning solutions to real-world problems.

Supervised students

Marco Latella, MSc at USI

2022

Graph Representation Learning for Multi-site Photovoltaic Energy Production

Talks

Reading group presentation at TGL reading group (Virtual)

Feb 2024

Invited to present the paper Taming Local Effects in Graph-based Spatiotemporal Forecasting at the Temporal Graph Learning Reading Group. (📹 Video)
Tutorial at ECML/PKDD (Turin)

Sep 2023

Presentation of the tutorial Graph Deep Learning for Spatiotemporal Time Series with colleagues Daniele Zambon and Andrea Cini.
Reading group presentation at TGL reading group (Virtual)

Apr 2023

Invited to present the paper Scalable Spatiotemporal Graph Neural Networks at the Temporal Graph Learning Reading Group.
Spotlight presentation at TGL Workshop (New Orleans)

Dec 2022

The Temporal Graph Learning Workshop at NeurIPS 2022.
Poster presentation at NeurIPS (New Orleans)

Nov 2022

The 36th Conference on Neural Information Processing Systems.
Invited talk at Baker Hughes (Virtual)

Jul 2022

Invited to give a webinar on time series imputation with colleague Andrea Cini.
Poster presentation at ICLR (Virtual)

Apr 2022

The 10th International Conference on Learning Representations.
Abstract presentation at MLDM (Virtual)

Nov 2021

The 10th Italian Workshop on Machine Learning and Data Mining.

Awards & Scholarships

Doctoral Mobility — Università della Svizzera italiana

Dec 2023

CHF 20’000 (≈$23K) to cover the expenses for a 6-month research stay at University of Oxford.
Travel Award — NeurIPS

Dec 2023

Travel award to attend the NeurIPS conference in New Orleans (US).
Best Paper Award — Temporal Graph Learning Workshop @ NeurIPS

Dec 2022

For the paper "Scalable Spatiotemporal Graph Neural Networks".
Travel Award — NeurIPS

Nov 2022

Travel award to attend the NeurIPS conference in New Orleans (US).
Scholarship — National Association SAPAR

2019

Scholarship awarded to the top-4 students in STEM subjects.
Scholarship — Politecnico di Milano

2019

Reduced tuition for high merits.

Projects

I believe in worldwide accessibility of science. As such, I make the software I develop for my research publicly available through my GitHub page. You can also find the code related to my publications on the GitHub page of Graph Machine Learning Group.

Torch Spatiotemporal

Torch Spatiotemporal (TSL) is a library built upon PyTorch and PyG for neural spatiotemporal data processing, with a focus on Graph Neural Networks.

GitHub Documentation

Other projects

GraPV

Sep 2020 — Feb 2023

Developing of graph-based methods for multi-site photovoltaic power forecasting, to improve accuracy on portfolio production prediction. The solution is based on novel graph-based AI strategies exploiting existing heterogeneous information and related dependencies. Joint project in collaboration with DXT Commodities, funded by Innosuisse.

Publications

Graph-based Forecasting with Missing Data through Spatiotemporal Downsampling
- Ivan Marisca ,
- Cesare Alippi ,
- Filippo Maria Bianchi
Preprint, 2024
Graph Deep Learning for Time Series Forecasting
- Andrea Cini ,
- Ivan Marisca ,
- Daniele Zambon ,
- Cesare Alippi
Preprint, 2023
Taming Local Effects in Graph-based Spatiotemporal Forecasting
- Andrea Cini ^* ,
- Ivan Marisca ^* ,
- Daniele Zambon ,
- Cesare Alippi
Advances in Neural Information Processing Systems, 2023
Scalable Spatiotemporal Graph Neural Networks
- Andrea Cini ^* ,
- Ivan Marisca ^* ,
- Filippo Maria Bianchi ,
- Cesare Alippi
Proceedings of the AAAI conference on artificial intelligence, 2023
Learning to Reconstruct Missing Data from Spatiotemporal Graphs with Sparse Observations
- Ivan Marisca ^* ,
- Andrea Cini ^* ,
- Cesare Alippi
Advances in Neural Information Processing Systems, 2022
Filling the G_ap_s: Multivariate Time Series Imputation by Graph Neural Networks
- Andrea Cini ^* ,
- Ivan Marisca ^* ,
- Cesare Alippi
International Conference on Learning Representations, 2022

^*Equal contribution.