Official implementation of the paper "Latent Functional Maps: a spectral framework for representation alignment." by Marco Fumero, Marco Pegoraro, Valentino Maiorca, Francesco Locatello and Emanuele Rodolà. Published at The Thirty-eighth Annual Conference on Neural Information Processing Systems, NeurIPS, 2024. This repository offers a lightweight, standalone implementation of the framework introduced in the paper.
Install the conda environment and dependencies via:
conda env create --name lfmap --file=lfmap.yml
conda activate lfmap
You're good to go!
You can explore basic functionalities and applications of the Latent Functional Map framework through the provided notebooks. These examples demonstrate its use in trasferring information, finding correspondences, and measuring similarity across various representational spaces, reproducing the paper’s results.
If you find our code or paper useful, please cite it at:
@inproceedings{NEURIPS2024_79be41d8,
author = {Fumero, Marco and Pegoraro, Marco and Maiorca, Valentino and Locatello, Francesco and Rodol\`{a}, Emanuele},
booktitle = {Advances in Neural Information Processing Systems},
editor = {A. Globerson and L. Mackey and D. Belgrave and A. Fan and U. Paquet and J. Tomczak and C. Zhang},
pages = {66178--66203},
publisher = {Curran Associates, Inc.},
title = {Latent Functional Maps: a spectral framework for representation alignment},
url = {https://proceedings.neurips.cc/paper_files/paper/2024/file/79be41d858841037987964e3f5caf76d-Paper-Conference.pdf},
volume = {37},
year = {2024}
}