Location: IDMC (Institut des Sciences du Digital), Nancy, France. How to get there
Date: 26 and 27 November, 2024.
This workshop will explore low-rank matrix and tensor decompositions/approximations and their interactions with neural networks and machine learning at large. The workshop will cover theoretical foundations as well as practical applications, with the main goal to connect researchers working in these fields. This workshop is also a closing event for the ANR LeaFleT project (project ANR-19-CE23-0021).
We will start at 13h-14h on 26 November and should finish on the 27th in the afternoon. Stay tuned for the detailed schedule of talks!
Registration is free but mandatory (the capacity is limited). Click here to register!
We will be announcing the full list of speakers shortly.
Overview of tensor decompositions and some applications to wireless communications
Study of a few properties of LLM pruning
Implicit Regularization in Regularized (Nonnegative) Low-Rank Approximations
Abstract: Regularized nonnegative low-rank approximations such as sparse Nonnegative Matrix Factorization or Sparse Nonnegative Tucker Decomposition are an important branch of dimensionality reduction models with enhanced interpretability. However, from a practical perspective, the choice of regularizers and regularization coefficients is often challenging because of the multifactor nature of these models and the lack of theory to back these choices. The work presented aims at improving upon these issues. By studying a more general model called the Homogeneous Regularized Scale-Invariant, we prove that the scale- invariance inherent to low-rank approximation models causes an implicit regularization with unexpected effects. This observation enables to better understand the effect of regularization functions in low-rank approximation models, to guide the choice of the regularization hyperparameters, and to design balancing strategies to enhance the convergence speed of dedicated optimization algorithms. We showcase our contributions on sparse Nonnegative Matrix Factorization, ridge-regularized Canonical Polyadic decomposition and sparse Nonnegative Tucker Decomposition.
Decoupling multivariate functions using tensors
Low rank approximation of moment matrices and tensors
On the multi-spiked random tensor model
Network compression using tensor decompositions and pruning
| Date | Time | Session |
|---|---|---|
| Tue 26 Nov | (Day 1) | |
| 13:45-14:00 | Registration | |
| 14:00-14:15 | Opening remarks | |
| 14:15-15:15 | André de Almeida | |
| 15:15-15:45 | Mariya Ishteva | |
| 15:45-16:15 | Coffee break | |
| 16:15-17:15 | Francesco Tudisco | |
| 17:15-17:30 | Spotlight posters | |
| 17:30-19:00 | Cocktail + posters | |
| Wed 26 Nov | (Day 2) | |
| 09:00-09:30 | Welcome coffee | |
| 09:30-10:30 | Bernard Mourrain | |
| 10:30-11:00 | Jeremy Cohen | |
| 11:00-11:30 | Coffee break | |
| 11:30-12:30 | Julia Gusak | |
| 12:30-14:00 | Lunch break | |
| 14:00-15:00 | Christophe Cerisara | |
| 15:00-15:30 | Invited talk | |
| 15:30-16:00 | Invited talk | |
| 16:00-17:00 | Farewell coffee break |
Organization committee:
You can email the organizers at: firstname.lastname@univ-lorraine.fr
© 2024 Workshop LoRaINNe