Section: New Software and Platforms

CoNFab: COnvolutional Neural FABric

Participants : Shreyas Saxena, Jakob Verbeek.

Despite the success of convolutional neural networks, selecting the optimal architecture for a given task remains an open problem. Instead of aiming to select a single optimal architecture, we propose Convolutional Neural Fabrics [20] that embed an exponentially large class of CNN architectures. The fabric consists of a 3D trellis that connects response maps at different layers, scales, and channels with a sparse homogeneous local connectivity pattern. The only hyper-parameters of the model (nr. of channels and layers) are not critical for performance. While individual CNN architectures can be recovered as paths in the trellis, the trellis can in addition ensemble all embedded architectures together, sharing their weights where their paths overlap. By the non-cyclic property of the trellis, its parameters can be efficiently learned using methods based on error back-propagation. The trellis parameters can be learned using standard methods based on back-propagation, at a cost that scales linearly in the fabric size. This software implements Convolutional Neural Fabrics by means of wrappers on top of the Caffe library to specify and learn such models.