Requests for Information (RFI) For Robust and Secure Machine Learning (ML)

Solicitation Number: RFI-AFRL-RIK-18-04 Notice Type: Special Notice Synopsis: Requests for Information (RFI) For Robust and Secure Machine Learning (ML) FEDERAL AGENCY NAME: Department of Air Force, Air Force Materiel Command, AFRL - Rome Research Site, AFRL/Information Directorate, 26 Electronic Parkway, Rome, NY, 13441-4514 1.0 GENERAL INFORMATION 1.1 THIS IS A REQUEST FOR INFORMATION (RFI) ONLY. This RFI is issued solely for information and planning purposes and does not constitute a Request for Proposal (RFP) or a promise to issue a RFP in the future. This request for information does not commit the Government to contract for any supply or service whatsoever. Further, the Air Force is not at this time seeking proposals and will not accept unsolicited proposals. Responders are advised that the U.S. Government will not pay for any information or administrative costs incurred in response to this RFI. All costs associated with responding to this RFI will be solely at the interested party's expense. Not responding to this RFI does not preclude participation in any future RFP, if any is issued. If a solicitation is released, it will be synopsized on the Federal Business Opportunities (FedBizOpps) website. It is the responsibility of the potential offerors to monitor this site for additional information pertaining to this requirement. 1.2 FEEDBACK. Submission of an abstract is voluntary. Respondents are advised that AFRL is under no obligation to provide feedback with respect to any information submitted under this RFI. 1.3 REGULATORY GUIDANCE. This publication constitutes a Request for Information (RFI) as defined in Federal Acquisition Regulation (FAR) 15.201(e), "RFIs may be used when the Government does not presently intend to award a contract, but wants to obtain price, delivery, other market information, or capabilities for planning purposes. Responses to these notices are not offers and cannot be accepted by the Government to form a binding contract." 2.0 REQUEST FOR INFORMATION (RFI) The Air Force Research Laboratory, Information Directorate (AFRL/RI) is seeking information to better understand existing vendor offerings and the landscape of research and development (R&D) towards robust and secure machine learning techniques. The Air Force is investigating robust and secure machine learning techniques to determine what algorithms, methods, and techniques will be necessary to ensure efficient and effective performance of current and future machine learning systems. Technical Challenge: Current state-of-the-art machine learning algorithms have been shown to be vulnerable to adversarial attacks which can cause decreased classification confidence or misclassification. Such attacks have been shown to be effective in both the "laboratory" setting, using digital manipulation of public datasets [1][2], and in the real-world, using physical manipulations of objects [3]. Defensive techniques have been proposed to resist these attacks, but are often quickly defeated by new adversarial methods. A greater understanding of ML model architectures is necessary to understand the root cause of these vulnerabilities, and inform intelligent design tradeoffs between efficiency and robustness. Therefore, we expect that solutions will require deep consideration of model training and inferencing architectures to enable maximum efficiency while retaining model robustness to adversarial inputs. Furthermore, we would like to evaluate the performances on relevant AF datasets across the domain of AFRL/RI research, to include video/image classification, communications, and cyber operations. [1] Szegedy, Christian, Wojciech Zaremba, Ilya Sutskever, Joan Bruna, Dumitru Erhan, Ian Goodfellow, and Rob Fergus. "Intriguing properties of neural networks." arXiv preprint arXiv:1312.6199 (2013). [2] Goodfellow, Ian J., Jonathon Shlens, and Christian Szegedy. "Explaining and harnessing adversarial examples." arXiv preprint arXiv:1412.6572 (2014). [3] Evtimov, Ivan, Kevin Eykholt, Earlence Fernandes, Tadayoshi Kohno, Bo Li, Atul Prakash, Amir Rahmati, and Dawn Song. "Robust physical-world attacks on machine learning models." arXiv preprint arXiv:1707.08945 (2017). 3.0 REQUEST FOR INFORMATION (RFI) ABSTRACTS 3.1 CONTENT NO CLASSIFIED INFORMATION SHOULD BE INCLUDED IN THE RFI RESPONSE. 3.2 FORMAT All abstracts shall state that they are submitted in response to this announcement. The abstracts will be formatted as follows: Section A: A cover page identifying the company or organization, street address, and the names, emails and telephone numbers of the point of contact. In the case of partnerships, please provide the appropriate information for the lead POC. Also provide a short summary statement of the company's or party's experience/capabilities and a short summary of the organization's experience in the areas described above. This section is not included in the page count. Section B: Technical Summary. The Government is assessing the current state-of-the-art and future robust and secure machine learning. The RFI responses should describe the product solution proposed, address coverage of the requirements stated in this RFI by the proposed solution, explain the potential advantage to the Air Force, and provide a rough order of magnitude for the cost of the proposed solution. The abstracts shall be limited to 8 pages. All abstracts shall be double spaced with a font no smaller than 12pt font. All responses to this announcement must be addressed to the POCs, as discussed in Section 6.0 of this announcement. Respondents are required to submit at least one electronic copy to the Government technical point of contact (TPOC) in Microsoft Office Word. 3.3 ADDITIONAL INFORMATION. The submitted documentation and content thereof becomes the property of the U.S. Government and will not be returned. No solicitation documents exist at this time. This is NOT an Invitation for Bid (IFB) or a Request for Proposal (RFP). The Government does not intend to award a contract on the basis of this request. This is a request for information announcement for planning purposes only. The Government will not reimburse costs associated with the documentation submitted under this request. Responders are solely responsible for all expenses associated with responding to this inquiry. Although proposal terminology may be used in this inquiry, your response will be treated as information only and will not be used as a proposal. This announcement is not to be construed as a formal solicitation. It does not commit the Government to reply to information received, or to later publish a solicitation, or to award a contract based on this information. 3.4 PROPRIETARY INFORMATION. This notice is part of Government market research. Information received as a result of this request will be considered as sensitive and will be protected as such. Any company or industry proprietary information contained in responses should be clearly marked as such, by paragraph, such that publicly releasable and proprietary information are clearly distinguished. Any proprietary information received in response to this request will be properly protected from unauthorized disclosure. The Government will not use proprietary information submitted from any one source to establish the capability and requirements for any future acquisition, so as to inadvertently restrict competition. 4.0 INDUSTRY DISCUSSIONS. AFRL/RI representatives may or may not choose to meet with potential offerors. Such discussions would only be intended to get further clarification of potential capability to meet the requirements. 5.0 SPECIAL CONSIDERATIONS. Multiple abstracts within the purview of this RFI announcement may be submitted by each responder. 7.0 SUBMISSION. RFI abstract due date is June 30, 2018. 8.0 AGENCY CONTACTS. Verification of government receipt or questions of a technical nature can be directed to the cognizant technical points of contact (TPOCs): Primary TPOC Mr. Ryan Luley Telephone: 315-330-3848 Email: ryan.luley@us.af.mil Questions of a contractual/business nature shall be directed to the cognizant Contracting Officer, as specified below: Gail Marsh Telephone: (315) 330-7518 Email: gail.marsh@us.af.mil