Deep Learning (DL) | Apply Online

Opportunity Summary:

This DARPA funding opportunity (DARPA BAA 09-40), titled "Deep Learning (DL)," seeks research proposals aimed at advancing deeply layered machine learning systems. The core goal is to develop what DARPA describes as a "universal machine learning engine": a single, general-purpose learning approach that operates across multiple internal layers (at least three) to transform raw data into increasingly higher-level and more useful representations. In practical terms, the program is pushing for learning systems that can automatically discover patterns, invariants, and correlations in data by building up complexity layer by layer, rather than relying on hand-crafted features or narrowly tuned methods for each problem.

A central requirement is broad applicability across different types of inputs, or modalities. The envisioned engine should work for multiple data sources (for example, imagery, audio, text, sensor streams, or other signals) with minimal changes beyond input preprocessing. That means the learning approach itself should not have to be reinvented each time the input format changes; instead, the same underlying multi-layer methods should generalize, with only front-end adjustments needed to convert each modality into an appropriate input representation.

Another major emphasis is learning from unlabeled data. DARPA expects the system to extract important structure and produce useful internal representations without needing labeled training examples. This implies an interest in self-supervised or unsupervised learning behaviors where the system can organize, compress, predict, or otherwise model the input data in a way that yields features or representations that become valuable for downstream tasks, even when explicit labels are scarce or unavailable.

The opportunity also calls out the need for sophisticated spatiotemporal representations, indicating that the program is not limited to static pattern recognition. Performer teams are expected to tackle data where both space and time matter, such as video, motion trajectories, multi-sensor tracking, or other time-evolving phenomena. The intent is that the deep learning engine can represent complex temporal dynamics and spatial structure, and then use those representations effectively for learning and inference.

From an administrative standpoint, this is a discretionary research and development opportunity run by DARPA's Information Processing Technology Office. The funding instruments listed include cooperative agreements, grants, other transaction-like mechanisms, and procurement contracts, giving DARPA flexibility in how it structures awards depending on the nature of the proposed work and the desired level of government involvement. There is no cost-sharing or matching requirement indicated.

Eligibility is listed as unrestricted, meaning the opportunity is broadly open to any type of applicant (such as universities, companies, nonprofits, research institutes, or other organizations), subject to any additional eligibility notes contained in the full announcement. The opportunity was posted on April 15, 2009, with an original and current closing date of April 14, 2010, and an archive date of May 14, 2010. The program is associated with CFDA number 12.910 (Research and Technology Development). For full announcement access, the posting references FedBizOpps (as it was used at the time), and the listed point of contact is Josh Alspector, Program Manager and BAA Coordinator, for assistance particularly if there are issues accessing the announcement electronically.

Browse more opportunities from the same agency: DARPA Information Processing Technology Office

Browse more opportunities from the same category: Science and Technology and other Research and Development

Next opportunity: State Fire Training Systems Grants

Previous opportunity: Microbicide Innovation Program (MIP V) (R21/R33)

USGrants.org Applicant Portal:

Are you interested in learning about about how to apply for this government funding opportunity? You can create a free applicant account and receive instant access to our applicant portal that many business owners like you have benefited from.

Apply for DARPA BAA 09 40

 

[Watch] how do funding administrators access proposals?

Applicants also applied for:

Applicants who have applied for this opportunity (DARPA BAA 09 40) also looked into and applied for these:

 

Grant application guides and resources

It is always free to apply for government grants. However the process may be very complex depending on the funding opportunity you are applying for. Let us help you!

Apply for Grants

 

 

Premium leads for funding administrators, grant writers, and loan issuers

Thousands of people visit our website for their funding needs every day. When a user creates a grant proposal and files for submission, we pass the information on to funding administrators, grant writers, and government loan issuers.

If you manage government grant programs, provide grant writing services, or issue personal or government loans, we can help you reach your audience.

Subscribe to Leads

 

Request more information:

Would you like to learn more about this funding opportunity, similar opportunities to "DARPA BAA 09 40", eligibility, application service, and/or application tips? Submit an inquiry below:

Don't forget to subscribe to our grant alerts mailing list to receive weekly alerts on new and updated grant funding opportunities like this one in your email.