Screening and Stabilizing Learned Image Manifolds

In 2017, DARPA's Information Innovation Office (I2O) incorporated Deep Learning Analytics' image manifold work into their "A DARPA Perspective on Artificial Intelligence." The video notes here (approved for public release) outline the process and illustrate how simple tSNE embeddings can capture image manifold properties (like look angle loops) in embeddings. If missing data gaps on data manifolds are easier to identify in lower dimensional embeddings, it may also inform research on deep learning methods to improve inference performance in domains with limited training data.

Featured Posts
Recent Posts
Search By Tags
No tags yet.
Follow Us
  • Facebook Basic Square
  • Twitter Basic Square
  • Google+ Social Icon



If your organization has data but doesn’t have the data science or machine learning expertise in house to extract all the value in it, get in touch with us. Machine learning and data science is what we do.

4601 N Fairfax Drive, Ste 1200

Arlington, VA, 22203

Main (571) 969-5638

Job Inquiries (202) 643-1882

Copyright 2020 General Dynamics Mission Systems, Inc.