SBIR/STTR Research Award

Neural Collapse in Infrared Search and Track Architectures for Directed Energy Applications

Award identifier 8fd2cd909510feb1a720fa58c54f47cfd7114cd78621e47a0e95085545002ed2Awarding agency Department of Defense · Air Force

Award amount

$179,999

Program and phase

STTR · Phase I

Award year

2025

Award details

Awarding agency: Department of Defense
Branch: Air Force
Recipient company: TOYON RESEARCH CORPORATION
Proposal award date: 2024-12-09
Contract end date: 2025-09-09
UEI: PK1CB3L39XX8
Solicitation number: 24.B
Topic code: AF24B-T002

Research abstract

Toyon Research Corporation and UC-Santa Barbara propose a research program to study deep neural network classifiers trained for directed energy applications. Trained to identify and classify objects in infrared image or video data, these models will be studied to determine whether they undergo neural collapse, whether the phenomenon can be induced, at what level of data volume and computational complexity collapse emerges, and what consequences there are to arriving at a collapsed state. Past work has identified a collapsed state with improved performance and robustness including improved generalizability of the learned feature embeddings. The aim of the proposed research is to extend these benefits to directed energy applications.

Technology keywords

Not reported

Other awards by this recipient

Verified by the same UEI. These are other public SBIR/STTR award pages, not a merged company record.

Awards with the same topic code

Verified by an exact topic code match. This is not a text-similarity match.

Research funding history

601 SBIR/STTR records share this UEI (1986–2026). Reported award amounts total $249,287,075. Reported amounts are not federal obligations.

Phases: Phase I, Phase II, Phase II – Direct to Phase II

Agencies: Department of Defense, Department of Commerce, Department of Homeland Security, National Aeronautics and Space Administration, Department of Transportation, Department of Energy