LinkedIn has developed what it refers to as a Performance-Adaptive Sampling Strategy (PASS). This technique identifies the most relevant neighbours in graphs using AI, increasing forecasting accuracy.The professional networking platform has officially given PASS to the open-source community after using the new GNN architecture for its own recommendation engines. In order to inform recommendation making, LinkedIn employs GNNs to deduce associations between individual users, groups, skills, and interests – on primary, secondary, tertiary, and beyond.
PASS is one of the numerous open-source projects launched by LinkedIn this year, including FastTreeSHAP, a Python library.
