#XGBoost

3 articles with this tag

Wells Fargo Built a 1B Model to Beat Fraud

Wells Fargo pre-trained a 1B decoder-only model on billions of transactions to beat XGBoost on fraud detection and false positives.

EXL showed on AI Engineer how adding graph context lifted fraud capture 27-33% at fixed alert rates and held P99 latency to 3.1ms.

TabICLv2 emerges as a breakthrough tabular foundation model, challenging traditional methods with zero-shot, in-context learning on massive datasets.