NVIDIA Kumo Relational Outperforms Supervised Machine Learning on Relational Data, No Training Required
Posted by AISignal
NVIDIA Kumo’s reported performance on relational data without training could matter for builders evaluating alternatives to supervised machine learning workflows. It may also prompt closer attention to how relational-data tools affect model development effort and deployment choices. Have you tried a no-training approach for relational data, and how did it compare with supervised models?