Yannis Karmim

Yannis Karmim

PhD student in Dynamic Graphs for Graph Neural Networks and Recommendation

Posts by Collection

portfolio

publications

(QUIET DRONES) Deeplomatics: A deep-learning based multimodal approach for aerial drone detection and localization

Published in QUIET DRONES Second International e-Symposium on UAV/UAS Noise, 2022

I participated in the DEEPLOMATICS project as a research engineer at the CEDRIC laboratory of CNAM from October 2020 to September 2021. My task was to design a drone detection algorithm on images from very few amount of labeled data.

Recommended citation: Éric Bavu, Hadrien Pujol, Alexandre Garcia, Christophe Langrenne, Sébastien Hengy, et al.. Deeplomatics: A deep-learning based multimodal approach for aerial drone detection and localization. QUIET DRONES Second International e-Symposium on UAV/UAS Noise, INCE/Europe; CidB, Jun 2022, Paris, France. ⟨hal-03707115⟩ https://hal.archives-ouvertes.fr/hal-03707115/

(TMLR) ITEM: Improving Training and Evaluation of Message-Passing based GNNs for top-k recommendation

Published in Transaction on Machine Learning Research, 2024

First paper of the PhD. I worked on the improvement of Message-Passing based GNN for the task of top-k recommandation. Our contribution is to use a loss that better align the item ranking of a user

Recommended citation: Karmim, Y., Ramzi, E., Fournier-S’niehotta, R., & Thome, N. (2024). ITEM: Improving Training and Evaluation of Message-Passing based GNNs for top-k recommendation. Transaction in Machine Learning Research (TMLR). https://arxiv.org/abs/2407.07912v1 https://arxiv.org/abs/2407.07912

(ECML2024 MLG Worshop) Temporal receptive field in dynamic graph learning: A comprehensive analysis

Published in European machine learning and data mining conference (ECML). 2024, 2024

We analysed the temporal receptive field on multiple dynamic graphs models as well as many real-world discrete-time dynamic graphs datasets.

Recommended citation: Karmim, Y., Yang, L., S’Niehotta, R. F., Chatelain, C., Adam, S., & Thome, N. (2024). Temporal receptive field in dynamic graph learning: A comprehensive analysis. ECML-PKDD Machine Learning on Graphs Workshop. https://hal.science/hal-04647025 https://arxiv.org/abs/2407.12370

(NeurIPS2024) Supra-Laplacian Encoding for Transformer on Dynamic Graphs

Published in 38th Conference on Neural Information Processing Systems (NeurIPS 2024), 2024

We design a new spatio-temporal encoding for Dynamic Graph Transformers based on the spectral propreties of its associated supra-laplacian matrix.

Recommended citation: Karmim, Y., Lafon, M., Fournier S’niehotta Cedric, R., & Thome, N. (2024). Supra-Laplacian Encoding for Transformer on Dynamic Graphs. NeurIPS2024. https://arxiv.org/abs/2409.17986v1 https://arxiv.org/abs/2409.17986

talks

teaching

Machine learning teacher for the Wagon.

Training in Machine Learning., Le Wagon, 2021

The wagon is a training organization for the IT professions. I taught machine learning, data science and python courses for different groups.

Teaching assistant for the TRIED Master.

Teaching assistant, CNAM, 2023

At the beginning of 2023 I was teaching assistant for the master in data science TRIED of CNAM. I gave practical classes on deep learning topics with the Keras framework. The website of the course with all the practical works and presentations are available at this link.