A Riemannian Network for SPD Matrix Learning

Authors: Zhiwu Huang, Luc Van Gool

AAAI 2017 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We show through experiments that the proposed SPD matrix network can be simply trained and outperform existing SPD matrix learning and state-of-the-art methods in three typical visual classification tasks.
Researcher Affiliation Academia Zhiwu Huang, Luc Van Gool Computer Vision Lab, ETH Zurich, Switzerland {zhiwu.huang, vangool}@vision.ee.ethz.ch
Pseudocode No No structured pseudocode or algorithm blocks were found.
Open Source Code No The paper mentions using source codes from authors for comparing methods ('For all of them, we use their source codes from authors with tuning their parameters according to the original works.'), but does not explicitly state that the source code for the proposed SPDNet is publicly available or provide a link.
Open Datasets Yes We use the popular Acted Facial Expression in Wild (AFEW) (Dhall et al. 2014) dataset for emotion recognition.
Dataset Splits Yes The database is divided into training, validation and test data sets where each video is classified into one of seven expressions.
Hardware Specification Yes For training the SPDNet, we just use an i7-2600K (3.40GHz) PC without any GPUs.
Software Dependencies No The paper does not provide specific software dependency details (e.g., library or solver names with version numbers) needed to replicate the experiment.
Experiment Setup Yes The learning rate λ is fixed as 10⁻², the batch size is set to 30, the weights are initialized as random semi-orthogonal matrices, and the rectification threshold ϵ is set to 10⁻⁴.