# Artificial Intelligence, Data Sciences, and Quantum Information Sciences

**R-2A activity** of project 1502 — C5ISR/Naval Space  
**Program element:** 0601153N — Defense Research Sciences  
**Component:** U.S. Navy · **Budget Activity:** 1  
**Vintage:** President's Budget PB2027  
**Canonical URL:** https://hitchintel.com/programs/0601153N/1502/a0  
**Parent:** https://hitchintel.com/programs/0601153N/1502

## Summary

This activity requests $19.4M in FY2027, 41% of project 1502, down 23% on FY2026. The R-2A exhibit describes it across FY2026–FY2027, including what the FY2027 money is planned to buy.

## What the FY2027 request buys

**FY2027 planned work.** Quantum Information Sciences: - Continue research of quantum states, devices, phenomena relative to the simulation, information processing and computing performance needs of naval systems. - Continue research on novel techniques for controlling quantum states to improve performance of information processors, sensors and clocks. - Continue research on demonstrations of systems having a quantum advantage in the solution of optimization problems and quantum simulation of complex physical systems. - Continue research exploring the distribution of entanglement in a quantum network and applications thereof. - Continue research on the use of single magnetic excitations for quantum information processing. - Continue research on protocols for efficient quantum error correction. - Initiate research on applications of distributed quantum information processing. Mathematical Data Science: - Continue basic research in mathematics, probability, statistics, signal processing, machine learning, data engineering, and information theory. - Continue efforts to develop advanced algorithms for analyzing massive datasets in real time, identify real patterns and avoid false positives. - Continue investigations regarding the development of advanced methods to integrate and extract common features from large heterogeneous domains. - Continue research investigations of privacy in complex networks. - Continue research efforts regarding the development of scalable reinforcement learning. - Continue research investigations of causal dependences in complex networks. - Continue research investigation regarding large language models (LLMs). - Continue research in reliability and trustworthiness of foundation models. - Continue research on evaluation of LLMs. - Continue research into training foundation models on small amounts of data. - Continue research into computationally efficient foundation models. - Continue research into emergent properties of foundation models. - Initiate development of new methods for reinforcement learning from offline datasets. - Initiate development of large language models that are robust to adversarial attacks. Machine Reasoning and Intelligence: - Continue developing the science base and computational methods for building versatile intelligent agents, which can function autonomously in uncertain, unstructured, uncontrolled, open-world environments, and can collaborate seamlessly with humans and other agents. - Continue basic research for developing robust computer vision systems, based on human vision, for automated understanding of surveillance imagery, perception for autonomous agents, and managing image/video libraries for after-action analysis and planning. - Continue basic research in machine self-learning for intelligent agents, inspired by human learning, for understanding real-world environments. - Continue basic research in learning and decision-making in multi-agent systems in dynamic, uncertain settings where there are many competitive and cooperative agents and information about intentions and rewards are not fully known. This research area has a wide range of applications in tactical and strategic planning, economic planning, etc. - Continue basic research in exploiting generative models, such as auto-encoders and diffusion models, to solve inverse problems. This research area has a wide range of applications in robot perception and automated understanding of surveillance imagery. - Continue basic research in developing rigorous mathematics, formal methods, and new analytical approaches for evaluating performance of deep generative models and predicting their performance. This research area has the potential for developing analytical tools for verification and validation of generative AI capabilities and for drastically reducing the need for empirical evaluation of such systems such that they can be deployed in safety critical applications. - Complete basic research in developing new mathematical methods for principled design of deep learning architectures and analysis of their behavior. This program is expected to develop techniques for predicting performance of learning-based systems, to improve their generalization abilities, and to reduce the need for empirical verification. -Initiate basic research in control and decision-making in networks of dynamic agents under uncertainties in perception and network models. This is of critical importance for robust and reliable prediction and control of network agents particularly in long-duration missions. Optimization and Discrete Mathematics: - Continue to identify exploitable mathematical structures within specific decision problems for the purpose of devising superior solution algorithms. - Continue investigations into methods for strategically formulating and solving optimization problems that arise in resource allocation, logistics, and system planning. - Continue investigations into new techniques that utilize convex optimization and duality theory to solve non-convex optimization problems. - Continue research on developing novel first-order methods for solving general classes of problems that include saddle point problems, problems with a large number of constraints, and machine learning problems. - Continue investigations into finding solutions to various forms of multi-agent, multi-round games. - Continue investigations into optimization algorithms that leverage potential future computational advances including quantum and pseudo-quantum computers. - Continue investigations into developing algorithms to solve models on Riemannian manifolds. - Continue investigations into developing algorithms that find optimal, distributionally robust policies for reinforced learning models. - Initiate investigations into models and algorithms for combinatorial problems with uncertainty inspired by unmanned autonomous vehicles. - Initiate investigations into the use of LLMs to interact with optimization models.

**FY2026 to FY2027 change.** The funding decrease from FY 2026 to FY 2027 is due to a funding realignment from PE 0601153N / PU 1502: C5ISR/Naval Space to PE 0601153N / PU 1501: Autonomy/AI.

## Before the request year

**FY2026 plans — current year.** Quantum Information Sciences: - Continue research of quantum states, devices, phenomena relative to the simulation, information processing and computing performance needs of naval systems. - Continue research on novel techniques for controlling quantum states to improve performance of information processors, sensors and clocks. - Continue research on demonstrations of systems having a quantum advantage in the solution of optimization problems and quantum simulation of complex physical systems. - Continue research exploring the distribution of entanglement in a quantum network and applications thereof. - Continue research on the use of single magnetic excitations for quantum information processing. - Complete research on the utilization of photonic and phononic devices for high performance quantum information processing. - Initiate research on protocols for efficient quantum error correction. Mathematical Data Science: - Continue basic research in mathematics, probability, statistics, signal processing, machine learning, data engineering, and information theory. - Continue efforts to develop advanced algorithms for analyzing massive datasets in real time, identify real patterns and avoid false positives. - Continue investigations regarding the development of advanced methods to integrate and extract common features from large heterogeneous domains. - Continue research investigations of privacy in complex networks. - Continue research efforts regarding the development of scalable reinforcement learning. - Continue research investigations of causal dependences in complex networks. - Continue research investigation regarding large language models (LLMs). - Initiate research in reliability and trustworthiness of foundation models. - Initiate research on evaluation of LLMs. - Initiate research into training foundation models on small amounts of data. - Initiate research into computationally efficient foundation models. - Initiate research in emergent properties of foundation models. Machine Reasoning and Intelligence: - Continue developing the science base and computational methods for building versatile intelligent agents, which can function autonomously in uncertain, unstructured, uncontrolled, open-world environments, and can collaborate seamlessly with humans and other agents. - Continue basic research in developing new mathematical methods for principled design of deep learning architectures and analysis of their behavior. This program is expected to develop techniques for predicting performance of learning-based systems, to improve their generalization abilities, and to reduce the need for empirical verification. - Continue basic research for developing robust computer vision systems, based on human vision, for automated understanding of surveillance imagery, perception for autonomous agents, and managing image/video libraries for after-action analysis and planning. - Continue basic research in machine self-learning for intelligent agents, inspired by human learning, for understanding real-world environments. - Continue basic research in learning and decision-making in multi-agent systems in dynamic, uncertain settings where there are many competitive and cooperative agents and information about intentions and rewards are not fully known. This research area has a wide range of applications in tactical and strategic planning, economic planning, etc. - Continue basic research in exploiting generative models, such as auto-encoders and diffusion models, to solve inverse problems. This research area has a wide range of applications in robot perception and automated understanding of surveillance imagery. - Initiate basic research in developing rigorous mathematics, formal methods, and new analytical approaches for evaluating performance of deep generative models and predicting their performance. This research area has the potential for developing analytical tools for verification and validation of generative AI capabilities and for drastically reducing the need for empirical evaluation of such systems such that they can be deployed in safety critical applications. Optimization and Discrete Mathematics: - Continue to identify exploitable mathematical structures within specific decision problems for the purpose of devising superior solution algorithms. - Continue investigations into methods for strategically formulating and solving optimization problems that arise in resource allocation, logistics, and system planning. - Continue investigations into new techniques that utilize convex optimization and duality theory to solve non-convex optimization problems. - Continue research on developing novel first-order methods for solving general classes of problems that include saddle point problems, problems with a large number of constraints, and machine learning problems. - Continue investigations into finding solutions to various forms of multi-agent, multi-round games. - Continue investigations into optimization algorithms that leverage potential future computational advances including quantum and pseudo-quantum computers. - Complete research on integrating machine-learning techniques with algorithms for stochastic and combinatorial optimization. - Complete investigations into applying topological data analysis to combinatorial optimization problems. - Initiate investigations into developing algorithms to solve models on Riemannian manifolds. - Initiate investigations into developing algorithms that find optimal, distributionally robust policies for reinforced learning models. Applied and Computational Mathematics: - Continue basic research in developing analytical and computational tools for models of physical phenomena of critical interest to the Navy in waves, flows, materials, structures and information processing. - Continue to develop robust, reliable and near-real-time computational models for predicting environmental behavior in atmospheric and oceanic processes.

## Funding

| Fiscal Year | Estimate Type | Amount ($M) |
|---|---|---|
| FY2025 | Actual | 0.0 |
| FY2026 | Enacted | 25.1 |
| FY2027 | Request | 19.4 |

> Prior, current and budget year only — an R-2A activity carries no five-year plan. It sums exactly into its project in the request year and not necessarily in any other.

## Other activities in project 1502

- Complex Software Systems and Cybersecurity — FY2027 7.9
- Communications and Networks — FY2027 7.2
- Information Technology — FY2027 4.9
- Quantum Measurement Architectural Devices — FY2027 3.6
- Electromagnetic Warfare — FY2027 2.1
- Research Platforms — FY2027 1.6
- Networked Sensing — FY2027 1.0
- Expeditionary Cyber — FY2027 0.0

## Source & machine access

- **Source:** FY2027 Department of the Navy RDT&E Budget Justification, Exhibit R-2A, PE 0601153N project 1502 (PB PB2027). Narrative is the government's own text.
- **No marks, no contractors at this grain** — congressional marks land on the program element and R-3 performers on the project.
- **MCP:** `mcp.hitchintel.com` — `budget_get_activity`.

*HitchAI is an independent intelligence service, not affiliated with the U.S. Department of Defense. Budget figures are requests/estimates, not obligations.*