Graph Neural Networks for Image Classification and Reinforcement Learning using Graph representations
This is an incremental study exploring graph neural networks in computer vision and reinforcement learning for researchers in those fields.
The paper tackled image classification by testing a novel non-redundant graph representation against trivial mappings, and reinforcement learning by modeling a Rubik's cube as a graph problem versus standard model-free methods, but no concrete results or numbers were reported.
In this paper, we will evaluate the performance of graph neural networks in two distinct domains: computer vision and reinforcement learning. In the computer vision section, we seek to learn whether a novel non-redundant representation for images as graphs can improve performance over trivial pixel to node mapping on a graph-level prediction graph, specifically image classification. For the reinforcement learning section, we seek to learn if explicitly modeling solving a Rubik's cube as a graph problem can improve performance over a standard model-free technique with no inductive bias.