In this project, Clifford algebra has been studied from a more operational perspective applied to neural networks. In the produced report, after recalling some mathematical notions to explain the foundations of Clifford algebra, tfga was introduced. This is a library written by Robin Kahlow to implement Clifford algebra with TensorFlow and thereby implement various layers used in neural networks, taking the latter into account.
We then reviewed several practical examples, culminating in the implementation of the ideas presented in the paper "Alberto Pepe, Joan Lasenby, and Sven Buchholz. CGAPoseNet+GCAN: A Geometric Clifford Algebra Network for Geometry-Aware Camera Pose Regression. 2024 IEEE/CVF Winter Conference on Applications of Computer Vision (WACV), Jan 2024" starting from a base provided by the authors of the paper, which originally only implemented CGAPoseNet.