Ameet Talwalkar, Assistant Professor of Computer Science, UCLA @ MLconf SF Part 1
summary
MLbase builds on Apache Spark and MLlib to automate model selection by searching the hyperparameter space instead of relying on ad hoc tuning. It combines new search algorithms, execution-level performance enhancements and statistical heuristics to prune trials and reuse work, reducing the number of full model fits. The system finds a deployed classifier faster and with far less compute, reporting roughly an order of magnitude speedup over standard grid- and random-search baselines.
AI-written summary · about
related talks
39mMachine Learning with Scala on Spark
38mFantastic ML apps and how to build them
26mMLbase: A Distributed Machine Learning System
44mApache Spark and ML workflows
01h 22mAutomated Machine Learning & Tuning with FLAML
11mAutomated ML: A journey from CRISPR.ML to Azure ML
50mApache Spark vs Machine Learning
17mMachine Learning at the Speed of Light with Spark