Moritz Böhle

Pytorch-LRP. Basic LRP implementation in PyTorch

B-cos. B-cos Networks: Alignment is All we Need for Interpretability

CoDA-Nets. Official implementation for the CVPR paper "Convolutional Dynamic Alignment Networks for Interpretable Classifications"

sfa-project. "The goal of this project is to study how Slow Feature Analysis (SFA) establishes spatial represen- tations similar to those found in the rodent hippocampus [Franzius et al., 2007]. Because visual data is very high-dimensional, we concentrate on a computationally simpler sensory system that is vaguely inspired by the echolocation system bats use to locate their prey. You will implement a simple movement pattern of a virtual ”bat” in a rectangular room, emulate its sensory input and apply SFA and independent component analysis to learn spatial representations."