SplitNet: Learning to Semantically Split Deep Networks for Parameter Reduction and Model Parallelization

81 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Split-and-Bridge: Adaptable Class Incremental Learning within a Single Neural Network (2021, methodology): “This idea is inspired by SplitNet (Kim et al. 2017), which is originally introduced to build a tree-structured network with multiple branches such that similar classes are assigned to the same part of the network parameters.”
3D Point Cloud Generative Adversarial Network Based on Tree Structured Graph Convolutions (2019, background): “Tree-structured Deep Networks: There have been several attempts to represent convolutional neural networks or long short-term memory using tree structures [5, 20, 27, 28, 31].”
Exploring Hidden Dimensions in Parallelizing Convolutional Neural Networks (2018, background): “Another common approach is model parallelism (Mirhoseini et al., 2017; Kim et al., 2017) that divides the network parameters into disjoint subsets and trains each subset on a dedicated device.”