Phase II year
2021
(last award dollars: 2023)
Phase II Amount
$2,686,875
BlueLightAI will build a suite of novel automated object detection and annotation tools for use on massive image data sets. Based on topological constructions that use our proprietary geometric feature technology, the BlueLightAI suite will incorporate expanded prior knowledge into innate detection, segmentation, identification, and annotation of images. Replacing current Convolutional Neural Networks with our proprietary Topological Neural Networks results in new deep learning algorithms and capabilities. The BlueLightAI tools will develop new data transforms and sparse architectures to deliver more reliable learning from much smaller amounts of data, with concomitant much smaller memory usage. In turn, this analysis will be more resistant to adversarial attacks, and achieve quantifiably greater accuracy in transfer and generalization across similar data sets. A substantial body of research initiated and conducted by the PI and his group has demonstrated proof of concept on a variety of examples and establishes feasibility for the pioneering BlueLightAI approach detailed here.
