Deep Compression of Sum-Product Networks on Tensor Networks
arXiv:1811.03963
Abstract
Sum-product networks (SPNs) represent an emerging class of neural networks with clear probabilistic semantics and superior inference speed over graphical models. This work reveals a strikingly intimate connection between SPNs and tensor networks, thus leading to a highly efficient representation that we call tensor SPNs (tSPNs). For the first time, through mapping an SPN onto a tSPN and employing novel optimization techniques, we demonstrate remarkable parameter compression with negligible loss in accuracy.
References in corpus (4)
- Tensor Decompositions for Signal Processing Applications From Two-way to Multiway Component Analysis
- Sum-Product Networks: A New Deep Architecture
- Analysis and Design of Convolutional Networks via Hierarchical Tensor Decompositions
- Learning Graph-Structured Sum-Product Networks for Probabilistic Semantic Maps