TECHNOLOGY AREA(S): Human Systems, Information Systems

ACQUISITION PROGRAM: Providentia FNC FY20 Candidate, MAGTF C2, MTC2

OBJECTIVE: Mature algorithms/software to enable assessments and predictive capabilities from graph data relevant to Naval use cases.

DESCRIPTION: Convolutional neural networks (CNNs) in recent years have revolutionized computer vision. Recurrent neural networks have enabled meaningful progress in natural language processing. Naval data, however, is to a significant extent graph based. For example, information about an opposing force (e.g., a platform or unit having some set of attributes/capabilities present at a specific location) is most effectively captured in a graph form. In the last couple of years, graphical convolutional networks have been developed with the goal of enabling CNN-based performance on images to translate to graph data.

Under this STTR topic, during Phase I awardees will tackle the problem of training a machine to make decisions from graph sequences [Ref 5] using open source data. Researchers will mature and code graphical convolutional neural networks to make predictions from dynamic text-derived graph data. Of particular interest is a capability that can predict future risk assessments and global trends in Global Database of Events, Language, and Tone (GDELT) from activity / content in Reddit.

During Phase II, awardees will train graph CNNs to make decisions on the capabilities and limitations and vulnerabilities/opportunities for units/forces from graphical data with the help of Government-furnished data. Awardees will be expected to mature their Phase I algorithms and train machines to assess units/forces based on their attributes/capabilities/conditions/locations. Specific technical challenges include algorithm development, feature selection, the design of a machine learning training plan, and results in good performance against test data.

Work produced in Phase II may become classified. Note: The prospective contractor(s) must be U.S. owned and operated with no foreign influence as defined by DoD 5220.22-M, National Industrial Security Program Operating Manual, unless acceptable mitigating procedures can and have been implemented and approved by the Defense Security Service (DSS). The selected contractor and/or subcontractor must be able to acquire and maintain a secret level facility and Personnel Security Clearances, in order to perform on advanced phases of this project as set forth by DSS and ONR in order to gain access to classified information pertaining to the national defense of the United States and its allies; this will be an inherent requirement. The selected company will be required to safeguard classified material IAW DoD 5220.22-M during the advanced phases of this contract.

PHASE I: Determine feasibility and complete a proof of concept study of the use of a graphical convolutional neural network for risk assessment and global trends. Conduct a detailed analysis of literature and commercial capabilities. For a bounded number of Reddit sub-groups and GDELT metrics, train a model that predicts one from the other. Carefully design a validation plan to verify performance of the resulting model. Develop a Phase II plan with a technology roadmap and milestones for generalizing the use of their algorithm.

PHASE II: Produce a prototype system based on the preliminary design from Phase I. Ensure that the capability of the prototype extends to a machine learning service that can predict the capabilities/limitations of a platform/force and suggest opportunities/vulnerabilities from graph-based data. Note: The system will need to ingest military graphical data at the secret level and provide explanatory evidence for unit/force assessments; and simulations may be needed to generate labeled data.� During Phase II, the small business may be given specific mission scenarios by the Government to validate capabilities. An awardee should assume that the prototype system will need to run as a distributed application with a mature design for the human computer interface. Deliver a working prototype of the system (source code and executable) and software documentation including a user�s manual, and provide a demonstration using a Naval operational scenario of interest.

It is probable that the work under this effort will be classified under Phase II (see Description section for details).

PHASE III DUAL USE APPLICATIONS: Produce a final prototype capable of deployment to training centers, operational command and control centers and as a virtual application. Adapt the system to transition as a component to a larger system or as a standalone commercial product. Provide a means for performance evaluation with metrics for analysis (e.g., accuracy of assessments) and a method for operator assessment of product interactions (e.g., display visualizations). The Phase III system should have an intuitive human computer interface. The software and hardware should be modified and documented in accordance with guidelines provided by the engaged Programs of Record and any commercial partners. Researchers are encouraged to publish S&T contributions.

Technology development should be applicable to any domain that requires assessments/predictions to be made from graph-based data including but not limited to common tactical picture assessment, readiness assessment, event detection, or video summarization.

REFERENCES:

1. Kipf, Thomas N. �Graph Convolutional Networks.� 30 September 2016. https://tkipf.github.io/graph-convolutional-networks/

2. Ganssle, Graham. �Node Classification by Graph Convolutional Network.� January 20, 2018. https://www.experoinc.com/post/node-classification-by-graph-convolutional-network

3. Kipf, Thomas N. and Willing, Max. �Semi-Supervised Classification with Graph Convolutional Networks.� ICLR 2017. https://arxiv.org/abs/1609.02907

4. Henaff, Mikael, Bruna, Joan, LeCun, Yann. �Deep Convolutional Networks on Graph Structured Data.� Submitted 16 June 2015. https://arxiv.org/abs/1506.05163

5.� Li, Yujia, et. Al, �Gated Graph Sequence Neural Networks� ICLR 2016 https://arxiv.org/abs/1511.05493v4

KEYWORDS: Graph Analysis; Fusion; Convolutional Neural Networks; Predictive Science; Natural Language Processing; Classifications