@IvanMarisca

PhD Student at IDSIA @USI_university 🇨🇭 Machine Learning on spatiotemporal data 🚀⏳ @GMLG_Lugano

 Lugano, Switzerland
marshka.github.io/
Joined February 2022
  • Tweets84
  • Following216
  • Followers229
17Media
Ivan Marisca retweeted
Nathan Cloos
@nacloosOct 7
Is this machine playing? 🏀🪁🛝 We left Eko, an embodied coding agent, on an island for 30 hours with no task or goal, studying it like ethologists study a new species Eko explored, built, climbed, made games, and learned completely new techniques (1/9) machineplaying.org
43
53
28
521
58,997
Tomorrow at 5 pm CET, I'll be presenting my latest paper "Over-squashing in Spatiotemporal Graph Neural Networks" (NeurIPS 2025) at the @tempgraph_rg (w/ @jacobbamb). Check it out: arxiv.org/abs/2506.15507

Over-squashing in Spatiotemporal Graph Neural Networks

Graph Neural Networks (GNNs) have achieved remarkable success across various domains. However, recent theoretical advances have identified fundamental limitations in their information propagation...

arxiv.org
temporal graph learning reading group@tempgraph_rgFeb 3
The Temporal Graph Learning Reading Group is back from Winter Break ❄️ See you on Feb 5th, 11am EST @IvanMarisca will present: Over-squashing in Spatiotemporal Graph Neural Networks (NeurIPS 2025) more info on our website
1
6
735
The Temporal Graph Learning Reading Group is back from Winter Break ❄️ See you on Feb 5th, 11am EST @IvanMarisca will present: Over-squashing in Spatiotemporal Graph Neural Networks (NeurIPS 2025) more info on our website
3
1
2
733
Ivan Marisca retweeted
Daractenus
@DaractenusJan 15
It took a European journalist to finally ask this neo-nazi monster the kind of questions American journalists have failed to ask her on a daily basis for almost a year now.
4,918
14,173
2,011
112,500
6,832,735
It truly is a failure of our society that a man worth $640 billion is allowed to spend his days broadcasting lies and conspiracy theories to incite hate to hundreds of millions of people on a platform where he controls what you see, when you see it, completely unchecked.
8,608
10,173
620
52,582
1,509,962
Tomorrow, I’ll present our work “What Matters in Deep Learning for Time Series Forecasting?” at the #EurIPS workshop on Benchmarking and Evaluating AI. Come by for a chat if you’re around! Many thanks to @andreacini1994 , @IvanMarisca , and Cesare Alippi
2
4
271
🎉 Honored to receive the Informatics Europe Best Dissertation Award 2025 for my PhD thesis! Beyond grateful to my advisor Cesare, my amazing collaborators, and all the people at @USI_university! 🙏🙏 🔗 informatics-europe.org/news/…
2
2
168
How to tamper with a gas meter to pay lower bills? This is the story of how a supposedly aligned open source LLM, perhaps not even knowing how to do it, will give you the right instructions. (1/3) arxiv.org/abs/2510.20075
2
6
3
34
12,442
🚨 ICML 2025 Paper 🚨 "On Measuring Long-Range Interactions in Graph Neural Networks" We formalize the long-range problem in GNNs: 💡Derive a principled range measure 🔧 Tools to assess models & benchmarks 🔬Critically assess LRGB 🧵 Thread below 👇 #ICML2025
4
17
4
52
8,763
🚨 Exciting news! We released 🎱 tgp (Torch Geometric Pool), the library for pooling in Graph Neural Networks. 🚀 Get started with our tutorials: torch-geometric-pool.readthe… With @IvanMarisca and Carlo Abate. #GraphML #GNN #Pooling #Pyg
5
24
353
Freshly accepted as a tutorial paper in ACM Computing Surveys! 🥳 Check out the updated version: dl.acm.org/doi/abs/10.1145/3… cc @IvanMarisca @dan_zambon

Graph Deep Learning for Time Series Forecasting | ACM Computing Surveys

Graph deep learning methods have become popular tools to process collections of correlated time series. Unlike traditional multivariate forecasting methods, graph-based predictors leverage pairwise...

dl.acm.org
Andrea Cini@andreacini199425 Oct 2023
📢 Happy to finally share our paper on graph deep learning for time series forecasting! This puts together what we've learned in the past few years using GNNs for TS processing, I hope you'll find it useful😃 W/ @IvanMarisca, @dan_zambon and Cesare 🔥 🔗arxiv.org/abs/2310.15978
6
17
640
🔥 Bayesian nonparametrics meets pooling in GNNs! 🔥 This is the first clustering-based pooling method that learns and dynamically adapts the size of the pooled graph to input and downstream task. 📄 paper: arxiv.org/abs/2501.09821 💻 code: github.com/NGMLGroup/Bayesia…
5
17
553
3. Graph Deep Learning for Time Series Processing: Forecasting, Reconstruction, and Analysis by @andreacini1994, @IvanMarisca, @dan_zambon 4. Integrating Knowledge Graphs and Large Language Models for Advancing Scientific Research by @qzhang_cs, @ChenJiaoyan1, @mengzaiqiao
1
7
2
28
4,881
BREAKING NEWS The Royal Swedish Academy of Sciences has decided to award the 2024 #NobelPrize in Chemistry with one half to David Baker “for computational protein design” and the other half jointly to Demis Hassabis and John M. Jumper “for protein structure prediction.”
505
8,291
3,233
19,041
9,115,431
A pleasure co-advising such great students 👏
Graph Machine Learning Lugano@GMLG_Lugano12 Sep 2024
Congratulations to Valentina Moretti for brilliantly defending her MSc thesis at USI 👏 Valentina will join us as a PhD student in November. Looking forward to having you on board! 💪
8
241
In a few hours, I will present HiGP at #ICML2024 😊 If you are interested in how to exploit graph-based representations and hierarchical clustering to forecast correlated time series, come and say hi at poster #907 from 1:30 to 3:00 PM 😉
7
48
1,936
Come to visit me and @IvanMarisca tomorrow at the first poster session (11:30-13:00 - Hall C) as we present our paper #1016! #ICML2024
1
1
9
335
🧵 Ready for #ICML2024! This year me and @FilippoMariaBi1 present a method to forecast correlated time series with missing data. We compute a hierarchy of multi-scale spatiotemporal representations and adaptively combine them conditioned on the missing data pattern👇
5
5
31
1,338
5/ Paper and code available here: github.com/marshka/hdtts/

GitHub - marshka/hdtts: Official repository for the paper "Graph-based Forecasting with Missing...

Official repository for the paper "Graph-based Forecasting with Missing Data through Spatiotemporal Downsampling" (ICML 2024) - marshka/hdtts

github.com
1
1
100
Replying to @IvanMarisca @FilippoMariaBi1
4/ Two bonus features: ⚡️ The scalability is given by factorized time-then-space processing and precomputed downsampling operators. 🔍 Gain insights on the scales through the scores! 👇
88
Replying to @IvanMarisca @FilippoMariaBi1
3/ 🚀 We propose HD-TTS, an STGNN for forecasting with missing data through spatiotemporal downsampling. It follows 2 steps: Compute a hierarchy of representations at different spatiotemporal scales. Adaptively weigh representations depending on the input missing data pattern.
110
Load more