A Lie Group Approach to Riemannian Batch Normalization

Authors: Ziheng Chen, Yue Song, Yunmei Liu, Nicu Sebe

ICLR 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Extensive experiments conducted on widely-used SPD benchmarks demonstrate the effectiveness of our framework.
Researcher Affiliation Academia 1 University of Trento, 2 University of Louisville
Pseudocode Yes Algorithm 1: Lie Group Batch Normalization (Lie BN) Algorithm
Open Source Code Yes The code is available at https://github.com/Git ZH-Chen/Lie BN.git.
Open Datasets Yes Radar dataset (Brooks et al., 2019b), HDM05 dataset (Müller et al., 2007), FPHA (Garcia-Hernando et al., 2018), Hinss2021 dataset (Hinss et al., 2021)
Dataset Splits Yes Ten-fold experiments on the Radar, HDM05, and FPHA datasets are carried out with randomized initialization and split (split is officially fixed for the FPHA dataset), while on the Hinss2021 dataset, models are fit and evaluated with a randomized leave 5% of the sessions (inter-session) or subjects (inter-subject) out cross-validation scheme.
Hardware Specification Yes All experiments use an Intel Core i9-7960X CPU with 32GB RAM and an NVIDIA Ge Force RTX 2080 Ti GPU.
Software Dependencies No The paper mentions software like PyTorch, MOABB, MNE, and geoopt, but does not provide specific version numbers for these dependencies.
Experiment Setup Yes The experiments are conducted with a learning rate of 5e 3, batch size of 30, and training epoch of 200 on the Radar, HDM05, and FPHA datasets. For the Hinss2021 dataset, following Kobler et al. (2022a), we use a learning rate of 1e 3 with a weight decay of 1e 4, a batch size of 50, and a training epoch of 50.