Lossless Compression of Structured Convolutional Models via Lifting
arXiv:2007.06567
Abstract
Lifting is an efficient technique to scale up graphical models generalized to relational domains by exploiting the underlying symmetries. Concurrently, neural models are continuously expanding from grid-like tensor data into structured representations, such as various attributed graphs and relational databases. To address the irregular structure of the data, the models typically extrapolate on the idea of convolution, effectively introducing parameter sharing in their, dynamically unfolded, computation graphs. The computation graphs themselves then reflect the symmetries of the underlying data, similarly to the lifted graphical models. Inspired by lifting, we introduce a simple and efficient technique to detect the symmetries and compress the neural models without loss of any information. We demonstrate through experiments that such compression can lead to significant speedups of structured convolutional models, such as various Graph Neural Networks, across various tasks, such as molecule classification and knowledge-base completion.
arXiv commentAccepted to ICLR 2021. TL;DR: Speeding up weight-sharing dynamic neural computation graphs, such as GNNs, with lifted inference
References in corpus (12)
- Semi-Supervised Classification with Graph Convolutional Networks
- Inductive Representation Learning on Large Graphs
- Neural Message Passing for Quantum Chemistry
- Fast Graph Representation Learning with PyTorch Geometric
- Representation Learning on Graphs with Jumping Knowledge Networks
- A Survey of Model Compression and Acceleration for Deep Neural Networks
- Deep Graph Library: A Graph-Centric, Highly-Performant Package for Graph Neural Networks
- How Powerful are Graph Neural Networks?
- Bisimulation-based Approximate Lifted Inference
- Understanding Isomorphism Bias in Graph Data Sets
- Synthesizing Datalog Programs Using Numerical Relaxation
- Neural Markov Logic Networks