Scalable Data Science & Deep Learning with H2O
summary
H2O is an in‑memory Java ML platform that compresses columns, ships code not data, and exposes R/Python/Scala/Java/JSON plus a Flow GUI; it parses large files (demo shows 160 million rows in 35s parsing into a 4GB compressed frame), builds a GLM in ~19s and reports model metrics, exports production scoring as POJO scoring code for real-time deploy, and runs distributed deep networks on 11M Higgs events reaching mid‑80s AUC; Spark integration appears as Sparkling Water.
AI-written summary · about
chapters (25)
- 0:00Intro
- 0:40Agenda
- 1:09H2O
- 3:04What is H2O
- 4:39H2O customers
- 5:15Downloading H2O
- 5:53Demo
- 10:04Metrics
- 17:03Deep Learning
- 21:27World Record
- 22:15Normal Digit Detection
- 35:25Crime Detection
- 35:52Datanami
- 36:41Spark Memory
- 37:44Parse Dataset
- 38:00Date Manipulation
- 38:27Spark
- 39:19Scala
- 40:55Flow
- 41:52State Up Table
- 42:17Python
- 42:57IPython Notebook
- 46:23Kaggle Challenge
- 47:12GBM
- 47:54Outlook
related talks
01h 10mGetting started with H2O on Python
45mMassive distributed processing with H2O
44mScalable Automatic Machine Learning with H2O
33mH2O internals
30mScalable Automatic Machine Learning with H2O
27mDeploying distributed AI and machine learning in financial services
32mDifferent Strategies of Scaling H2O Machine Learning on Apache Spark
01h 10mIntroduction to Machine Learning with H2O and Python