AutoML

AutoML is a platform designed to assist researchers in the process of hyper-parameter tuning by intelligently searching for the optimal parameter settings to train models automatically.

• Clara AutoML high-level design
  
  • AutoML as a platform
  • Use cases
  • Search space and parameter search
  • Controller and executor
  • AutoML workflow
  • Handlers and workflow events
  • AutoML with Clara Train
• AutoML user guide
  
  • Specify search space
  • Configuring the “search” section
  • Command Line Interface (CLI) for AutoML
• AutoML programming guide
  
  • Introduction
  • AutoML Context
  • SearchSpace
  • Parameter Range Locator (PRL)
  • Search result and recommendations
  • Outcome - result of recommendation execution
  • Understanding AutoML workflow
  • Write your own controller
  • Write handlers
  • Write your own executor
  • Create your own AutoML system
• AutoML search space definition for Clara Train
  
  • Components in Clara Train
  • Defining the search space