Snowflake: Scaling GNNs to High-Dimensional Continuous Control via Parameter Freezing

by   Charlie Blake, et al.

Recent research has shown that Graph Neural Networks (GNNs) can learn policies for locomotion control that are as effective as a typical multi-layer perceptron (MLP), with superior transfer and multi-task performance (Wang et al., 2018; Huang et al., 2020). Results have so far been limited to training on small agents, with the performance of GNNs deteriorating rapidly as the number of sensors and actuators grows. A key motivation for the use of GNNs in the supervised learning setting is their applicability to large graphs, but this benefit has not yet been realised for locomotion control. We identify the weakness with a common GNN architecture that causes this poor scaling: overfitting in the MLPs within the network that encode, decode, and propagate messages. To combat this, we introduce Snowflake, a GNN training method for high-dimensional continuous control that freezes parameters in parts of the network that suffer from overfitting. Snowflake significantly boosts the performance of GNNs for locomotion control on large agents, now matching the performance of MLPs, and with superior transfer properties.


Evasion Attacks to Graph Neural Networks via Influence Function

Graph neural networks (GNNs) have achieved state-of-the-art performance ...

GNN-Ensemble: Towards Random Decision Graph Neural Networks

Graph Neural Networks (GNNs) have enjoyed wide spread applications in gr...

Equivariant Polynomials for Graph Neural Networks

Graph Neural Networks (GNN) are inherently limited in their expressive p...

Transfer Learning of Graph Neural Networks with Ego-graph Information Maximization

Graph neural networks (GNNs) have been shown with superior performance i...

Shift-Robust GNNs: Overcoming the Limitations of Localized Graph Training data

There has been a recent surge of interest in designing Graph Neural Netw...

From Graph Low-Rank Global Attention to 2-FWL Approximation

Graph Neural Networks (GNNs) are known to have an expressive power bound...

Please sign up or login with your details

Forgot password? Click here to reset