Doing More With Less: Accelerating the Analyst From Modeling Down to the Hardware

Abstract

The goal of our project is to build systems that make it easier than before for analysts to extract information from a wide variety of data sources and then use this information for analytical tasks, including prediction, regression, and ever sophisticated modeling. We have developed two systems. A TA2 system for improving model search by using developer exhaust and a TA3 system called snorkel for generating training sets using weal supervision. The snorkel system has been spectacularly successful, makes building high quality ML models much easier for a variety of problem domains. Snorkel has been used extensively by academia, government and industry.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Aug 01, 2020
Accession Number
AD1108186

Entities

People

  • Oyekunle A. Olukotun

Organizations

  • Stanford University

Tags

Communities of Interest

  • Autonomy
  • Cyber

DTIC Thesaurus Topics

  • Air Force
  • Air Force Research Laboratories
  • Artificial Intelligence Software
  • Computational Science
  • Computer Programming
  • Computers
  • Data Mining
  • Governments
  • Information Processing
  • Information Science
  • Information Systems
  • Machine Learning
  • Monte Carlo Method
  • Natural Language Processing
  • Neural Networks
  • Supervision
  • Training

Fields of Study

  • Computer science

Readers

  • Agent-Based Social Robotics and Mobile-Assisted Learning in Virtual Environments.
  • Distributed Systems and Data Platform Development
  • Neural Network Machine Learning.