Google Cloud AutoML Connected System

Overview

Google Cloud AutoML enables building, training, and predicting on custom Image, Text, or Translate models in using Google’s AutoML service. Model details include metadata about the type of model.

Key Features & Functionality

Integrations provided:

• Create Dataset
  • Create a new Image, Text, or Translate Dataset
• Create Model
  • Trains a new Model from a Dataset (note: can take many hours)
• Delete Dataset
  • Deletes a dataset and its data
• Export Data
  • Exports data from dataset to new location in Google Cloud Storage
• Get Dataset
  • Returns a Dataset’s Metadata
• Get Model Evaluation
  • Returns a Model’s Performance evaluation
• Get Model
  • Returns a Model’s Metadata
• Import Data
  
  • Imports data into Dataset from Google Cloud Storage
• List Datasets
  • List all datasets in project
• List Model Evaluations
  • List Model Evaluations for a specific model
• List Models
  • List Trained Models in project
• Predict
  • Uses a specific model to Classify an image or block of text, or translate the text to another language
• Get Long Running Operation Status
  • Long-running operations now return immediately with their long-running operation name

Requirements:

• A Google Service Account with access to AutoML https://cloud.google.com/automl/ and Cloud Storage https://cloud.google.com/storage/.
• Highly recommended: The Google Cloud Storage Connected System Plug-in for pushing data to Cloud Storage to populate the dataset.