In-database machine learning: Automate, scale, deploy
summary
Oracle shows in-database AutoML that picks algorithms, tunes hyperparameters, and deploys models without moving data: a GUI plus OML4Py runs automated algorithm selection (example result: svm_gaussian 98%), feature selection that shrinks 13 predictors to 9, and model tuning. Models (30+ in-database algorithms) scale near-linearly from 100M to 800M rows, support autoscaling, and export to OML services or ONNX for REST scoring with prediction details and explainability.
AI-written summary · about
chapters (16)
- 0:00Introduction
- 0:32Data volumes are growing exponentially
- 1:59Indatabase machine learning
- 4:52Overview
- 5:50Modeling process
- 6:35Traditional ML pain points
- 7:47Oracle ML AutoML UI
- 8:40Demo
- 11:45Oracle ML for Python
- 12:28OML for Python
- 16:23Automl for SQL
- 17:40Scalability
- 19:58Deployable using productionready infrastructure
- 23:22Modeling and productionready deployment
- 26:22Questions to consider
- 27:25Summary
related talks
26mOracle Machine Learning: what's new and what's next?
22mPredict the future (or don't) with machine learning in applications
54mUnderstand & Where to use AI & Machine Learning 101
28mAutoML—hype vs. value
23mModern database monitoring with integrated machine learning
59mAlgorithmia in Action: Take your ML models from training to production
17mBuild GenAI Apps with Oracle Database 23ai
44mScalable Automatic Machine Learning with H2O