Sissy Kosma

✅This allows CNNs to learn representations robust to common time series distortions, such as offset shift and linear trend. These invariant representations improve the network’s robustness across tasks like classification, anomaly detection, and transfer learning. (4/n)

✏️To address this, we propose a mathematical framework that uses group actions for time series to model how certain deformations affect time series data. Subsequently, we propose deformation-free representations of time series. (3/n)

🚀 Our paper "Time Series Representations with Hard-Coded Invariances", with Thibaut Germain & Laurent Oudre, is accepted to @icmlconf! 🔗 Paper: openreview.net/pdf?id=SaKPKyj… 📍If you're attending #ICML2025, pass by poster#E-1906 on 15 July (Session 2, East)! (1/n)

Excited to share that our paper, “The Signed Two‐Space Proximity Model for Learning Representations in Protein–Protein Interaction Networks,” with @chkosma, @Brativnychka, @MichailChatzia1, @IakovosEvd, and @mvazirg has been accepted at OUP Bioinformatics @OUPBioinfo! (1/n)

🎉 Excited to announce our paper "Signed Graph Autoencoder for Explainable and Polarization-Aware Network Embeddings" with @nnaknik @giannis_nikole @chkosma @IakovosEvd @mvazirg has been accepted at #AISTATS2025! 🚀🧵👇 (1/n)

In Vienna for #ICLR2024! Meet us w/ @georgepanago5 📌Halle B#29 (Thu Morn.) We present our TMLR paper (w/@mvazirg,@giannis_nikole,JM.Steyaert): iclr.cc/virtual/2024/p… - We present a Neural ODE-based model for improved generalization in predicting epidemic spreading on graphs.

It is experimentally shown that our approach remains robust and computationally effective when tested on several real-world networks and experimental settings. - Featured Certification, Expert Certification

In this paper, we focus on the SIR epidemic model for networks and propose a novel Neural ODE-based method. The proposed method employs a message passing mechanism similar to GNNs and incorporates knowledge of the epidemic process through an approximated ODE system.

Excited that our paper entitled “Neural Ordinary Differential Equations for Modeling Epidemic Spreading” with @giannis_nikole, @georgepanago5, Jean-Marc Steyaert and @mvazirg, has been accepted for publication at the TMLR journal. twitter.com/TmlrPub/status…

Hello, world. My first look at my forever home. #CountdownToMars

"Girls shouldn't play with cars", "Women are not usually good at maths", "they are less pragmatic" etc. Women in science have never thought that their sex can stop them from learning. Or they may haven't been left to. And that's something to celebrate every day. #WomenInSTEM

@ChrisPanag thank you 😀

@ChrysoulaKosma Welcome to Twitter! 😀