Fig. 13: Illustration of the topological gating mechanism on a mini-graph.
From: Geometric multi-instance learning for weakly supervised gastric cancer segmentation
This schematic depicts a simplified six-node example with two clusters representing normal (pastel blue) and tumor (pastel red) tissue regions. The central node i is highlighted with a learnable gate σ(gi), which adaptively controls the blending between the node’s intrinsic feature hi and its neighborhood descriptor si.
