Research Blog

2018

A new parallel feedback architecture

3 minute read

The forward-pass classification of the Deep Convolutional Networks (DCNs) inhibit them to actually understand the scene represented by the image. It is restricted to understand each image as lines and edges.