Jure Leskovec, Professor of Pc Science at Stanford College and Chief Scientist at Kumo.ai, speaks with host Sriram Panyam about relational and graph language fashions and their transformative influence on enterprise decision-making and predictive modeling.
Jure begins by establishing the important significance of predictive modeling throughout industries – from fraud detection in monetary establishments to buyer churn prediction, lifetime worth estimation, product suggestions, and healthcare threat evaluation. He notes that whereas AI has made outstanding advances in pure language understanding and laptop imaginative and prescient, predictive modeling over enterprise operational information saved in relational databases has been largely left behind, nonetheless counting on 30-year-old machine studying approaches which are costly, time-consuming, and require guide characteristic engineering.
His proposed resolution to the basic downside with present approaches is relational deep studying and relational transformers. The dialogue explores how this method differs from conventional graph neural networks (GNNs), which Jure pioneered and deployed efficiently at Pinterest. Jure concludes with sensible steerage for software program engineers and information scientists enthusiastic about exploring this expertise.
Delivered to you by IEEE Pc Society and IEEE Software program journal.
