AFOSR - Information and Networks

INFORMATION AND NETWORKS (RTA2)

The Information and Networks Team within the Engineering and Information Science Branch is organized to support many U.S. Air Force and Space Force priority areas including autonomy, space situational awareness, and cyber security. The research programs within this team lead the discovery and development of foundational issues in mathematical, information and network-oriented sciences. They are organized along three (3) themes: Information, Decision Making, and Networks.

The information theme addresses the critical challenges faced by the U.S. Air Force and Space Force which lie at the intersection of the ability to collect, mathematically analyze, and disseminate large quantities of information in a time critical fashion with assurances of operation and security.

Closely aligned with the mathematical analysis of information is the need for autonomous decision making. Research in this theme focuses on the discovery of mathematical laws, foundational scientific principles, and new, reliable and robust algorithms, which underlie intelligent, mixed human-machine decision-making to achieve accurate real-time projection of expertise and knowledge into and out of the battle space.

Information analysis and decision making rarely occur in the context of a single source. The networks theme addresses critical issues involving how the organization and interaction among large collections of information providers and consumers contributes to an understanding of the dynamics of complex information systems.

The Information and Networks (AFOSR/RTA2) Program Officers and topics are:

  • Computational Cognition and Machine Intelligence
  • Computational Mathematics
  • Dynamical Systems and Control Theory
  • Dynamic Data and Information Processing
  • Information Assurance and Cybersecurity
  • Mathematical Optimization
  • Science of Information, Computation, Learning, and Fusion
  • Trust and Influence
  • Complex Networks
  • Cognitive and Computational Neuroscience
Our research areas of interest are described in detail below:
 

COMPUTATIONAL COGNITION AND MACHINE INTELLIGENCE



Program Description: This program supports innovative basic research on the fundamental principles and methodologies needed to enable intelligent machine behavior, particularly in support of mixed-initiative (i.e., human-machine teaming) systems. The overall vision of this program is that future computational systems will achieve high levels of performance, adaptation, flexibility, self-repair, and other forms of intelligent behavior in the complex, uncertain, adversarial, and highly dynamic environments faced by the U.S. Air Force and Space Force. This program covers the full spectrum of computational and machine intelligence, from cognitively plausible reasoning processes that are responsible for human performance in complex problem solving and decision-making tasks, to non- cognitive computational models of intelligence necessary to create robust intelligent systems. Robustness in this context is the ability to achieve high performance given at least some or all of the following factors: uncertainty, incompleteness or errors in knowledge; limitations on sensing; real-world complexity and dynamic change; adversarial factors; unexpected events including system faults; and out-of-scope requirements on system behavior. In the midst of this spectrum are the technologies explicitly needed to seamlessly incorporate intelligent computational systems into mixed human-machine teams. The program is divided into three (3) sub-areas that span the full spectrum of computational and machine intelligence. They are: Computational Cognition, Human-Machine Teaming and Machine Intelligence.

The program encourages cross-disciplinary teams with collaboration including computer scientists, neuroscientists, cognitive scientists, mathematicians, statisticians, operation and management science researchers, information scientists, econometricians and game theoreticians, etc., especially when the research pertains to common issues and when collaboration is likely to generate bidirectional benefits. This program is aggressive, accepts risk, and seeks to be a pathfinder for U.S. Air Force and Space Force research in this area.

Proposals that may lead to breakthroughs or highly disruptive results are especially encouraged.

Basic Research Objectives: The Computational Cognition sub-area supports innovative basic research on high-order cognitive processes that are responsible for good human performance in complex problem solving and decision-making tasks – we only want to model the things people excel at. The sub-area also seeks to support research on building computational systems that derive from and/or integrate cognitive and biological models of human and animal intelligence. The overall objective is to understand and exploit these processes to create computational models that perform as well as or better than the reasoning systems they emulate. This sub-area seeks basic research that pertains to exploiting the capabilities of the mind and brain (human or animal) for creating more intelligent machines, as well as cognitively plausible mechanisms inspired by human (or animal) reasoning. This includes computational models based on human and animal performance in perception, attention, memory, learning, reasoning, and decision making in order to improve machine performance.

This sub-area does NOT, however, support statistical approaches to machine learning (e.g., “Deep Learning”), or related variants, as fundamental science in that area is addressed by the Science of Information, Computation, Fusion and Learning program described elsewhere in this BAA.

The Machine Intelligence sub-area supports innovative basic research on fundamental principles and methodologies of computational intelligence necessary to create robust intelligent systems. These methodologies may be cognitively inspired, or non- cognitive in nature, taking full advantage of the strengths embodied in mathematical and computational systems, such as the ability to reason with complex formal logic. This sub- area encourages research enabling the creation of computational systems that embody intelligent behavior based on cognitively inspired or purely mathematical approaches.

Proposals that lead to advances in the basic principles of machine intelligence for memory, reasoning, planning, scheduling, and cognitively-inspired learning (i.e., NOT “Deep Learning” or other statistical means), action, and communication are desired insofar as these contribute directly towards robustness as defined above.

The Human-Machine Teaming sub-area is primarily concerned with the machine-side of mixed human-machine decision-making, which appears at all levels of U.S. Air Force and Space Force operations and pervades every stage of U.S. Air Force and Space Force missions. To that end, new theoretical and empirical guidance is needed to prescribe maximally effective mixtures of human and machine decision making in environments that are becoming increasingly complex and demanding as a result of the high uncertainty, complexity, time urgency, and rapidly changing nature of military missions. This sub-area seeks new empirical and theoretical basic research that enables intelligent machines to perform as true “teammates,” adapting their behavior to accommodate changes in the environment, as well as augmenting the performance of human teammates when needed. This includes basic science in collaborative human- machine teams to aid the machine-side of inference, analysis, prediction, planning, scheduling, and decision making.

You are highly encouraged to contact our Program Officer prior to developing a full proposal to briefly discuss the current state-of-the-art, how your research would advance it, the approximate cost for a three (3) to five (5) year effort, and if there are any specific submission target dates.

DR. BEN ROBINSON, AFOSR/RTA2
Email: machine.itel@us.af.mil

COMPUTATIONAL MATHEMATICS

Program Description: This program seeks to develop innovative mathematical methods and fast, reliable and scalable algorithms aimed at making radical advances in computational science and large-scale engineering and design. Research in computational mathematics underpins the fundamental understanding of complex physical phenomena and leads to predictive simulation capabilities that are crucial to the design and control of future U.S. Air Force and Space Force systems, and their lifetime expectancy.

Proposals to this program should focus on fundamental scientific and mathematical innovations and should have the potential to address some of the most important computational challenges in science and engineering. Additionally, it is desirable to frame the basic research ideas in the context of applications relevant to the U.S. Air Force and Space Force, which can serve simultaneously to focus the research and to provide avenues for transition of basic research outcomes into practice. Applications of current Air Force and Space Force interest include, but are not limited to, quantum physics and quantum information systems, plasma dynamics, turbulence (e.g., in fluids, combustion, plasma), lasers and directed energy, aero-thermo-dynamics, information science, data analysis (including machine learning), biophysics, and material and structural sciences.

Basic Research Objectives: Research under this program has traditionally emphasized schemes that address the discretization and numerical solution of complex systems of equations, generally partial differential equations derived from physical models. However, alternative computational approaches are of keen interest, particularly in connection with emerging and multidisciplinary applications. Increased emphasis in this portfolio is placed on approaches that can handle a very high number of dimensions, uncertainty and stochasticity for non-Markovian processes, far from equilibrium conditions, and/or a wide range of scales (space, time, physical parameters, or complexity).

Research areas of particular interest currently include:
 
  • Mathematical methods for complexity reduction of high-dimensional, non- linear and multiscale problems, e.g., via projection-based methods and/or new machine- learning concepts. Such systems may have continuous, discrete or mixed representations, and may reside on graphs with evolving topology.
  • Mathematical approaches to the modeling of non-equilibrium statistical processes and turbulent dynamics with multiple physical interactions and large parameter spaces; of special interest are methods which effectively allow bi- directional transfer of information across scales, and can simultaneously reduce the computational burden while preserving the correct physics of interaction, including conservation laws and instability regimes.
  • Highly efficient and accurate methods for high-dimensional, nonlinear and stochastic dynamics with constraints. In particular, we are seeking revolutionary approaches to solving Hamilton-Jacobi-Bellman equations, optimal transport problems, and inverse problems for highly complex conditions. Of particular interest are applications in large-scale game theory, self-organized criticality and cascades, and the prediction of rare and extreme events.
  • Traditional computational methods involving high-order spatial and temporal algorithms remain of interest, if they have the potential for significant breakthrough and are able to meet the formidable computational challenges associated with current and future engineering problems of interest to the U.S. Air Force and Space Force.
The list above is not exhaustive and other approaches can be suggested to the Program Officer, who can then determine if a proposal is warranted and of potential interest. All proposed methods must be innovative, have quantifiable measures of fidelity, efficiency and adaptively, must be based on rigorous analysis and preferably demonstrated on canonical challenge and grand challenge problems.

You are encouraged to contact our Program Officer prior to developing a full proposal to briefly discuss the current state-of-the-art, how your research would advance it, the approximate cost for a three (3) to five (5) year effort, and if there are any specific submission target dates.

DR. FARIBA FAHROO, AFOSR/RTA2
Email: comp.math@us.af.mil

DYNAMICAL SYSTEMS AND CONTROL THEORY

Program Description: This program emphasizes the interplay of open dynamical systems and control theories, with the aim of developing innovative synergistic strategies for the design and analysis of controlled systems that enable radically enhanced capabilities, including performance and operational efficiency for future U.S. Air Force and Space Force systems. Only high-risk, and high-impact basic research on the mathematics of analysis and control theory for complex dynamical systems will be considered. Proposals should focus on the fundamental science and mathematics, while having relevance to future Air Force and Space Force systems and operations.

These applications currently include, but are not limited to, cyber/physical systems, biological and neurological systems, quantum systems, as well as autonomous/semi- autonomous aerial vehicles, munitions, and space vehicles. Note that research specifically restricted to an Air Force and Space Force application may not be considered fundamental research; therefore, future Air Force and Space Force applications should only be mentioned as examples or possibly motivations for fundamental theory in dynamical systems and control theory.

The dramatic increase in complexity of Air Force and Space Force systems provides unique challenges for the Dynamical Systems and Control Theory Program. Meeting these challenges may require interdisciplinary approaches as well as deeper studies within single disciplines.

Lastly, note that the Dynamical Systems and Control Theory Program places special emphasis on mathematically rigorous techniques addressing realistic treatment of applications, complexity management, autonomous and semi-autonomous systems, and real-time operation in stochastic and adversarial environments.

Basic Research Objectives: Current research interests include: algebraic, topological, and geometric methods of dynamical analysis of controlled complex systems for the purpose of real-time control; control of ensemble and infinite dimensional systems including PDE control; deterministic time and/or real-time reachability ; considering constraints, uncertain, information rich, dynamically changing, networked environments with time-varying topologies; robust and adaptive non-equilibrium control of nonlinear processes where the primary objective is enhanced operability rather than just local stability; new methods for understanding and mitigating the effects of uncertainties in dynamical processes where uncertainty distribution is non-Gaussian; novel theory for control of hybrid systems that can intelligently manage actuator, sensor, and processor communications in a complex, spatially distributed and evolving system of systems; sensor rich, data driven adaptive control; and applying control concepts motivated by studies of biological systems. Developments from nontraditional areas of applied mathematics such as applied algebra, geometry, and topology for the benefit of analysis and control synthesis of complex dynamical systems is also of interest. Furthermore, mathematical theory that is also applicable to more expressive representations of complex dynamics systems (e.g., hyper- and multi-graphs rather than graphs) is also welcome.

In general, the control of large complex, multi-scale, hybrid, highly uncertain nonlinear systems is of increasing interest. Furthermore, new mathematics in clear support of dynamics and control is of fundamental importance to this portfolio. In this regard, some areas of interest include, but are not limited to, hybrid dynamical systems theory, geometric and algebraic methods of dynamics and control, stochastic and adversarial systems, control of cyber physical systems with formal specifications, emerging areas of control theory, graph theoretic control theory over nonlinear dynamics at nodes of graphs, partial and corrupted information, nonlinear control and estimation, and novel computational techniques specifically aimed at control of systems with large data.

You are highly encouraged to contact our Program Officer prior to developing a full proposal to briefly discuss the current state-of-the-art, how your research would advance it, and the approximate cost for a three (3) to five (5) year effort.

DR. FREDERICK LEVE, AFOSR/RTA2
Email: dycontrol@us.af.mil

DYNAMIC DATA AND INFORMATION PROCESSING

Program Description: The portfolio seeks mathematical concepts that dynamically incorporate additional data, whether measured or from models, into an executing application to coordinate dynamic measurement collections, model refinements, and system awareness. Examples include dynamic data driven applications systems (DDDAS), physics-enhanced machine learning (PEML), and physics-based and human-derived information fusion (PHIF). Key developments should harness the use of first-principle models towards real, simulated, or augmented signals, data, and information processing to obtain substantial comparative improvements over existing methods. The portfolio encourages multidisciplinary research, especially synergistic and systematic collaborations between domain researchers in engineering, mathematics, computer sciences for multiresolution systems modeling, diagnostics, and analytics.

Basic Research Objectives: Foster individual and multidisciplinary research, technology development, and system analysis over emerging science and technology frontiers.

Domain Modeling: Methods leveraging large-scale simulations for real-time control, in concert with heterogeneous data collection, model updates, and system processing. Research advances should describe different levels of detail and modalities, invoke appropriate models, and interface with other data systems. For example, ideas to engender an integration of large-scale simulations, models, and data to advance traditional data processing paradigms.

Mathematical and Statistical Algorithms: Design methods for stable and robust convergence properties under perturbations induced by time-dependent (periodic and non-periodic, scheduled and event-driven) data inputs, multiple scales and model variations. Address enhanced asynchronous algorithms with stable communication between networked resources, multimodal modeling, and uncertainty quantification. For example, concepts to dynamically invoke models requiring elegant methods of uncertainty quantification, management, and propagation.

Measurement Systems and Methods: Innovate instrumentation platforms for collecting data, registering measurements, controlling sampling rates, and multiplexing multisource information. For example, designs to determine heterogeneous and embedded distributed sensor networks architectures, information fusion paradigms, and operationally robust performance.

Areas of interest to the Air Force and Space Force include:
 
  • Autonomy (e.g., leveraging large-scale modeling of mission planning, collaborative/cooperative control, and data learning for data analytics);
  • Agility (e.g., designing computational methods of sensor-based processing, ad-hoc network configurations, and multi-scale multi-physics simulations for decision support);
  • Authority (e.g., coupling high-performance aircraft health monitoring, space situational awareness, and ground operations for command and control);
  • Activeness (e.g., understanding materials stresses and degradation; embedded diagnostics, complex adaptive systems verification and validation, and cognitive performance augmentation for situational understanding).
New approaches are encouraged for combining computational, empirical, cognitive, theoretical, and analytical methods for interactive testing of multiple scientific and engineering hypotheses. Programmatic activities launched under this initiative will support research in individual areas of interest to the Air Force Research Laboratory, but mostly in the context of multidisciplinary research across the Basic Area Objectives mentioned above.

You are highly encouraged to contact the Program Officer prior to developing a full proposal to discuss the current state-of-the-art, how your research would advance it, the approximate cost for a two (2) to three (3) year effort, and if there are any specific submission target dates.

DR. ERIK BLASCH, AFOSR/RTA2
Email: D2IP@us.af.mil

INFORMATION ASSURANCE AND CYBERSECURITY

Program Description: Securing cyberspace, i.e. defending against and preventing cyber-attacks are not new challenges but these have become increasingly pressing in the light of technological advancements. Software and protocols are continuously becoming more complex to meet application demands. More flexible computing environments, such as distributed systems, demand new ways of thinking how to ensure secure end-to-end functionalities, even though components are only known to be individually secure. The emergence of nanoscale devices and quantum information processing and communication also portends new technological challenges for cybersecurity. By the same token, these new technologies potentially offer unparalleled security solutions to the existing or future problems.

Although engineering practices continue to provide short-term and temporary relieves to these pressing needs, new scientific ideas are required to address the lack of security and the explosive growth of hostile actions in cyberspace, especially taking into account of emerging technologies. Many fundamental concepts are still eluding precise formulation and awaiting rigorous responses. The goal of this Basic Research program is to explore novel, promising concepts and methodologies that can establish a firm scientific foundation for cybersecurity and potentially tackle the difficult technical hurdles described above.

Basic Research Objectives: Recent developments and advances in the following research areas of computer science and mathematics are expected to provide valuable insights into various cybersecurity problems: dependent type theory, cryptographic protocols for interactive computation and communication, interactive and automated theorem proving, language-based techniques in software and hardware for formal specification and verification, secure protocols, game theory with strong security content, obfuscation and fully homomorphic encryption, model categories, formalized mathematics. Broadly speaking, cross-fertilization of mathematical formalisms and logical constructs will likely continue to play a central role in the construction and verification of security invariants, and in the study of security models or security principles.

These scientific advances are expected to contribute fresh ideas to a number of fundamental cybersecurity topics: composition of security properties and protocols in distributed interactive systems without the need of trusted third parties; rigorous techniques to enable persistent and secure operations on unsecure or untrusted systems; information flow security and non-interference in dynamic and distributed settings; new security invariants that can readily be computed and interpreted, especially for systems endowed with rich geometric dynamics; rigorous proofs and construction of obfuscation techniques for programs and circuits to enhance security; formal verification and certification of the correctness of complex large-scale mathematical proofs and critical computer systems.

Aside from software and secure protocols, nanoscale material properties and quantum effects should offer added security capabilities for future computing devices that cannot be realized by today’s technologies. They potentially enable physical construction of cryptographic primitives that are traditionally described by algorithms and typically implemented by software. Random Number Generators and Physical Unclonable Functions are the simplest examples of such construction. At the same time, securing future unconventional technologies will require the introduction of new security principles and security models that may substantially deviate from the traditional approaches. In fact, various concepts in quantum information science and quantum computation such as quantum resources (entanglement, non-locality, contextuality, etc.) and quantum computational/communication complexity are highly relevant to the security of future communication and computing systems in which classical and quantum devices interact.

Research areas of interest to this program include, but are not limited to, the methodologies and topics described above. Highest priority will be given to projects with novel scientific ideas that potentially deliver new DoD/Air Force and Space Force capabilities.

You are highly encouraged to contact the Program Officer prior to developing a full proposal to briefly discuss the current state-of-the-art, how your research would advance it, the approximate cost for a three (3) to five (5) year effort, and if there are any specific submission target dates.

QUANTUM SOFTWARE RESEARCH PROGRAM

Program Description: With the advent of industrial quantum computing platforms, there have now been demonstrations of platforms with hundreds [2], or even one thousand [1] qubits suitable for executing gate-based quantum algorithms. Recent experiments using elements of error correction [9, 10] or error mitigation methods [3, 4, 5] have allowed circuits using tens or hundreds of qubits to be executed successfully even in the presence of noise.

However, mapping quantum circuits to hardware efficiently remains a difficult problem. Depending on the available gate set and connectivity of the underlying hardware, the depth of compiled circuits can vary by 3x or more, while the time to compile can vary by orders of magnitude [12, 13]. Small-scale compilers have been developed that are able to map circuits efficiently, but many do not scale to the hundreds or thousands of qubits presently available in commercial systems [11-13]. Moreover, the functionality of current compilers tends to be focused on a single platform [6, 7], without cross-platform functionality, and compilers are often formulated in terms of physical-level level gates with limited opportunities for abstraction.

The present state of quantum compilation stands in contrast to the classical compiling ecosystem. There, robust abstraction layers allow algorithms to be formulated in terms of primitives and basic operations, which can be efficiently mapped to hardware. Key to this is the creation of a compiler ecosystem, with multiple interchangeable compilation components fused together to form a stack. This allows both transparency, wherein the user need not tailor code for the specific processor in use, as well as diverse community of contributors to contribute compatible optimizations.

We are seeking proposals for research in quantum compilation and quantum software targeted at the creation of a compiler ecosystem. Successful proposals should aim to create a foundation for a unified ecosystem for compiling a range of algorithms across many platforms, such as superconducting qubits, spin qubits, photonics systems, and neutral atoms/trapped ions, as well as address the technical areas defined below.

TA1: Formal Compiler Architecture. Definition of layers and interfaces in a mathematical framework that allows for unit tests and debugging/verification. This includes:
 
  • A formal definition of all intermediate representations (IRs)
  • A mathematical theory of the consistency of these representations, translations and transpilations between them, and benchmarks for their performance on different architectures.
  • Methods to verify the performance of individual elements or layers, including unit testing, randomized and probabilistic testing, assertions, or formal verification schemes.
  • Example data structures
TA2: Quantum Computer Architecture. Data structures for defining architecture elements, including memories and connectivity. This includes:
 
  • Class definitions for high-level abstractions such as quantum memories or Q-RAMs, classical inputs connectivity or I/O bottlenecks
  • Lower-level abstractions such as modular architectures and topology.
  • Enumeration of utility to formal compiler elements.
TA3: Compiler Elements. Seeding initial versions of compiler elements, for example routers or circuit synthesizers, and setting up a framework for open source contributions. This includes:
 
  • The packaging of existing open-source elements into code that can interface with the rest of the stack, or production of new elements.
  • Documentation for external contributors to add packages to the framework
  • Responsibility for encouraging community contributions to the architecture.
A central metric for this research shall be compiled circuit depth and time-to-compile for a standard benchmark battery [8] on a variety of preprogrammed topologies. Proposing teams shall select an existing benchmark battery, and propose multiple topologies which form reasonable abstractions of multiple common platforms.

Out of scope:
 
  • Testing of specific algorithms, rather than the selected benchmark battery.
  • Costing or resource estimation for algorithms.
  • Error correction (all qubits should be assumed to be perfect)
  • Optimization of gates on physical qubits.
  • Unitary synthesis
Proposals should present a plan for addressing at least two of the three (3) TAs, and include all required expertise such as including classical compilation, programming languages, and quantum computing. Open-source release of software developed is encouraged.
 
  1. https://research.ibm.com/blog/quantum-roadmap-2033
  2. https://arxiv.org/abs/2306.11727
  3. https://journals.aps.org/prl/abstract/10.1103/PhysRevLett.119.180509
  4. https://arxiv.org/abs/2201.09866
  5. https://journals.aps.org/rmp/abstract/10.1103/RevModPhys.95.045005
  6. https://arxiv.org/abs/2306.03487
  7. https://arxiv.org/pdf/2207.01964.pdf
  8. For example, QASMBench or [], but other batteries are acceptable as well.
  9. https://arxiv.org/abs/2404.02280
  10. Bluvstein et al Nature volume 626, pages 58–65 (2024)
  11. https://arxiv.org/pdf/2306.03487.pdf
  12. https://arXiv.org/pdf/2202.14025v1.pdf
  13. https://arxiv.org/abs/2306.08629
Applicants should submit proposals under the Information Assurance and Cybersecurity portfolio and include “Quantum Software” at the beginning of the submission title.

DR. TRISTAN N. NGUYEN, AFOSR/RTA2
Email: info.security@us.af.mil

MATHEMATICAL OPTIMIZATION

Program Description: The program goal is to develop novel theory, algorithms, and software for the many classes of Mathematical Optimization problems that arise in support of decision, design, and allocation problems confronting the U.S. Air Force and Space Force. Areas of interest include resource allotment, planning, logistics, interdiction, engineering design, resiliency, and scheduling. Problems can be deterministic in that input parameters and objectives are known with certainty or can have data uncertainty that is addressed using such methods as stochastic programming and robust optimization. The research, while of fundamental importance to such problems, can also profoundly impact related areas of study, including the operation of autonomous vehicles and the effectiveness of machine learning.

Basic Research Objectives: The program welcomes basic research in theory, algorithms, and computational methods for continuous and discrete problems, both deterministic and stochastic. Contributions/impacts can be generally applicable to large families of problems or be suitably tailored to exploit specific mathematical structures found within special, important classes. As basic research aimed at having the broadest possible impact, the development of computational methods should include an emphasis on theoretical underpinnings, on rigorous convergence analysis, and on establishing performance guarantees for approximation methods.

Areas of interest include, but are not limited to:
 
  • Integer and mixed-integer programming
  • Continuous, nonconvex optimization
  • Multi-level optimization
  • Conic programming
  • Combinatorial optimization
Stochastic programming methodologies include:
 
  • Cutting plane and polyhedral methods for mixed-integer programs
  • Decomposition methods for large, specially structured problem
  • Global optimization for nonconvex programs
  • Interior-point and first- and second order methods for conic/convex/nonconvex optimization
The application process is as follows. Submit a white paper of length at most four pages to the Program Officer that summarizes the research to be performed, the importance, the approach, a summary of senior personnel qualifications, and an estimate of the total cost. If desired, additional pages can describe further qualifications and include a bibliography.

The white paper should not be overly technical but instead provide a clear description of the problem(s) to be solved, the approach and novelty, and the potential application to DoD missions. Then, depending on such factors as reviews and available budget, select white papers will be chosen for full-proposal development. A key evaluation criterion is the identification of innovative idea(s) that show promise for advancing the field of mathematical optimization, with reference to the optimization/mathematical programming literature.

Investigators are encouraged to contact the Program Officer prior to white paper submission to discuss the proposed research and how it will advance the state of the art, the approximate cost, and any specific target dates.

DR. WARREN ADAMS, AFOSR/RTA2
Email: optimization@us.af.mil

SCIENCE OF INFORMATION, COMPUTATION, LEARNING AND FUSION

Program Description: The U.S. Air Force and Space Force collects vast amounts of data through various modes at various times in order to extract and derive needed “information” from these large and heterogeneous (mixed types) data sets. Some data, such as those collected from magnetometers, register limited information content which is more identifiable at the sensor level but beyond human’s sensory reception. Other types of data, such as video cameras or text reports, possess more semantic information that is closer to human cognition and understanding. Nevertheless, these are instances of disparate data which encapsulate different types of “information” pertained to, perhaps, the same event(s) captured by different modalities through sensing and collection.

In order to understand and interpret information contained in various data sources, it is necessary to extract relevant pieces of information from these datasets and to make inferences based on prior knowledge and probabilities. This bottom-up processing direction needs conceptually driven reasoning to integrate or fuse the previously extracted snippets of information by leveraging domain knowledge. Furthermore, the top-down processes can offer causal explanation or causal inference, generate new hypotheses, verify or test hypotheses in light of observed datasets. Between the data- driven and conceptually-driven ends, there may reside different levels of abstraction in which information is partially extracted and aggregated based on the nature of applications.

Basic Research Objectives: With the rationale and guiding principles outlined in the above paragraph, this program seeks fundamental research that potentially leads to scientific advancements in informatics, computation, and learning that can support processing and making sense of complex disparate information sources. After all, information processing can formally and fundamentally be described as computing and reasoning on various knowledge representations. Successes in addressing the research sub-areas stated below would give the U.S. Air Force and Space Force new capabilities to: (1) shift emphasis from sensing to information awareness; (2) understand the underpinning of autonomy; (3) relieve human’s cognitive overload in dealing with the data deluge problem; (4) enhance human-machine interface in information processing.

To accomplish the research objectives, this program focuses on, but is not limited to, new techniques in mathematics, computing science, statistics and logic which have potentials to: (1) cope with various complex disparate data/information types; (2) integrate a diversity of unique reasoning and learning components collaborating simultaneously (e.g., multi-strategy reasoning and learning); (3) bridge correlational with causal discovery; (4) determine solutions or obstructions to local-to-global data- fusion problems; (5) mechanize reasoning/learning and computing in the same computational environment; (6) yield provably efficient procedures to enable or facilitate data analytics; (7) deal with high-dimensional and massive datasets with provably guaranteed performance.

You are highly encouraged to contact our Program Officer prior to developing a full proposal to briefly discuss the current state-of-the-art, how your research would advance it, and the approximate cost for a three (3) to five (5) year effort, and if there are any specific submission target dates.

DR. RICHARD D. (DOUG) RIECKEN, AFOSR/RTA2
Email: icf@us.af.mil

TRUST AND INFLUENCE

Program Description: The Trust and Influence program funds interdisciplinary high risk, transformative basic research that (1) elucidates the social and cognitive principles and processes surrounding the establishment, maintenance, and repair of trust between and among humans and intelligent agents, machines, algorithms, and/or other emergent technologies, with particular interest in situations where these concepts apply to heterogeneous, distributed teams or teaming constellations (i.e. teams of teams); and (2) advances the science of social influence to enhance understanding of how the phenomena and/or associated processes shape or affect human beliefs, perceptions, attitudes, and/or behaviors. The program encourages multidisciplinary and transdisciplinary approaches, which may include contributions from sociology, anthropology, computer and information science, psychology, cognitive science, linguistics, mathematics, economics, computational social science, and other social or behavioral sciences, among other disciplines. It further encourages research designs that utilize laboratory studies, modeling, and/or field research intended to develop novel, transformative theories, frameworks, or evaluative measures. Projects that aim to successfully consider the interplay of trust and influence in any of the contexts described below are welcomed.

Basic Research Objectives: The program’s research themes can be defined broadly by two areas: trust in autonomous systems, human-machine teaming, and emergent technologies; and influence. In the area of trust in autonomous systems, human- machine teaming, and emergent technologies, there is particular interest in research that covers one or more of the following:
 
  1. empirical studies to examine drivers of trust between/among humans and intelligent, autonomous agents that operate in various configurations of heterogeneous, distributed teams or constellations;
  2. laboratory and field studies to examine the impact of socially-designed and/or culturally-specific cues or physical features such as appearance, voice, personality, and other social elements related to human-machine trust and enhancing system performance;
  3. development of trust metrics, evaluative frameworks, and other relevant constructs in human-machine or human-agent teaming with a particular emphasis on real-time evolving and dynamic assessment;
  4. modeling and measuring of human-machine or human-autonomy teaming or joint action that supports adaptive and continuous improvement of joint performance in complex environments; and/or
  5. efforts that advance the state of the science surrounding AI and ethics, and/or responsible AI (RAI) in regard to human-technology interaction/teaming and trust, with a preference for research examining understudied ethical principles and theories;
  6. New constructs or theories of trust that may be required when considering non-maleficence and/or psychological safety.
In the area of influence, research is needed to understand how vectors of information, such as but not limited to social and digital media channels, contribute to short- and long-term campaigns intended to influence populations, shape perceptions, spread ideas, alter or diffuse narratives, and change beliefs. Further, there is interest in projects that consider multiple vectors of information that strategically influence beliefs and actions at multiple levels, layers, and scales. The portfolio is concerned with sociobehavioral, technosocial/sociotechnical effects and the interplay between cognitive and affective processes that give rise to behavior and actions, as well as those that contribute to how populations either become more susceptible or resilient to influence over time and how this occurs. The program does not fund projects that examine disinformation unless they are appropriately situated within the larger information ecosystem and/or multi-vector informational influence campaign(s). Specific areas of interest include research that considers one or more of the following:
 
  1. effects of influence and persuasion that occur within the information environment through multiple vectors of information (e.g. via campaigning), such as social, digital, and analog media, and across different cultural groups, with consideration of the associated second and third order effects;
  2. social/sociocultural, cognitive, and neural mechanisms of influence and persuasion;
  3. new or revised theories of resonance, to include research that considers it in conjunction with associated concepts such as dissonance, alienation, and others that might be revealed through investigation;
  4. modeling and measurement of the relationship between online and offline real-world behaviors; and/or
  5. novel measurements of social influence and persuasion, with an interest in efforts that consider multiple levels and scales.
Proposers are highly encouraged to contact the Program Officer prior to developing a full proposal, preferably by email, to discuss: the current state of understanding in the field(s); how the research would advance it; proposed or potential research questions or hypotheses; definitions of major concepts, such as trust and/or influence; the approximate cost for a three (3) to five (5) year effort; and specific submission target dates. If a white paper is available, it should be included with the initial email.

DR. LAURA STECKMAN, AFOSR/RTA2
Email: ti@us.af.mil

COMPLEX NETWORKS

Program Description: Complex networks are pervasive in military, commercial, and civilian operations. Complex networks consist of a graph (directed or undirected) together with a set of attributes. These attributes can include scalar or multi-dimensional weights on the edges or nodes of the graph, topological characteristics of the graph, flows over the graph, and processes that define the dynamics of the graph. Complex networks cut across many scientific disciplines (e.g., mathematics, computer science, engineering, socio-economics, etc.) and many application domains (e.g., communications, sensing, information systems, transportation, etc.). Networks fundamentally describe the structural aspects of interactions between individual agents. Networks can be extremely large and can have multiple characteristic scales. They can be static or dynamic. They can be physical or virtual. Networks can consist of multiple heterogeneous subnetworks (i.e., a network- of-networks), with explicit and implicit interdependencies. For example, transportation networks are intimately coupled to computer and electrical-power networks. Thus, the failure of a critical node or arc in one network can trigger failures in another, which can create a cascade event with catastrophic consequences. All of these characteristics of networks can make the analysis, understanding, and utilization of networks difficult and computationally prohibitive.

This basic-research program is focused on developing fundamental mathematical and algorithmic techniques to design, analyze, utilize, and understand complex networks. The program seeks innovative approaches with far-reaching potential, meaning any mathematical and algorithmic approaches ideally should be applicable to broad classes of problems and not tied to a particular application domain. The networks of interest can have arbitrary topologies, can be static or dynamic, and can be subject to uncertain conditions, ranging from a stochastic environment to deliberate adversarial actions affecting both nodes and links.

Basic Research Objectives: This program is aimed at developing mathematical and algorithmic tools for the design, analysis, understanding, and utilization of complex networks. Problems of interest include, but are not limited to:
 
  • Network resilience and robustness: This includes techniques for defining and measuring the resilience of a dynamic network; techniques for predicting, identifying, and mitigating adversarial actions against a network; techniques for network interdiction.
  • Network analytics: This includes algorithmic techniques for solving important classes of problems on networks, such as optimal resource allocation and information dissemination; algorithmic techniques for inference problems on networks that extract global information about network structure and function from local information. The algorithms can be exact or approximate with performance guarantees. Of particular interest are decentralized algorithms and understanding the tradeoff between centralization and decentralization.
  • Complexity reduction of networks: This includes techniques for decomposition, sparsification, and dimensionality reduction of networks with the goal of making networks easier to analyze, understand, and visualize; techniques for analyzing and exploiting graph-theoretic structure such as multi-layer networks.
  • Extension of network-analysis techniques to more general combinatorial, algebraic,or analytical structures: This includes network-relevant research in submodular functions, hypergraphs, matroids, graphons, etc.
Approaches may draw upon, but are not limited to, techniques from graph theory, optimization, complexity theory, algorithmic game theory, combinatorics, linear algebra, statistics, and probability.

You are highly encouraged to contact the Program Officer, preferably by email, prior to developing a full proposal, to briefly discuss the current state-of-the-art, how your research would advance it, the approximate cost, and if there are any specific submission target dates.

DR. DONALD K. WAGNER, AFOSR/RTA2
Email: complex-networks@us.af.mil

COGNITIVE AND COMPUTATIONAL NEUROSCIENCE

Program Description: The Cognitive and Computational Neuroscience program funds high-risk, high-potential basic research that uses experimental and computational modeling techniques from systems neuroscience, cognitive neuroscience, computational/theoretical neuroscience, cellular and molecular neuroscience, cognitive science, and cognitive psychology to:
 
  1. understand the neural mechanisms responsible for perception, cognition, and behavior/motor control.
The program also supports:
 
  1. brain-inspired algorithm and hardware development provided these are useful for testing proposed neuroscience theories and/or enabling novel capabilities in computing, artificial intelligence, or autonomous systems.
Basic Research Objectives: The program funds basic research in cognitive and computational neuroscience and at the intersection of neuroscience and artificial intelligence as described above. Potential topics that would be of interest to the program include but are not limited to the following examples. Innovative ideas that may not fit into these categories but are germane to the program’s general objectives are also welcomed.
 
  • Neural Information Representation. It is well established that neurons communicate information via action potentials (“spikes”), but understanding how neurons represent information has remained a long-standing challenge. Debates continue regarding whether neurons use precise spike timing or frequency to encode information, the functions of noisy, probabilistic population codes, and whether every spike carries signal. Proposed research projects should characterize neural activity with the aim of reliably decoding information. Proposers should describe how their research, if successful, will advance our ability to test hypotheses regarding neural mechanisms and functions and/or enable greater sophistication for applications such as sensory and sensorimotor prosthetics, brain-machine interfaces, and deception detection.
  • Bio-inspired Sensing. To process sensory information, recognize and locate objects, and navigate in real-world environments, humans and other animals often use strategies that outperform current man-made systems. Artificial systems featuring such strategies could augment human capabilities while enhancing or maintaining situational awareness. Proposals should describe a species’ sensory function in which the sensor, environment, and behavior align with the proposed application(s), hypotheses about the associated neural information processing mechanisms, and a research plan that culminates in a proof-of-concept demonstration.
  • Brain-inspired Machine Learning. Current machine learning algorithms excel at identifying statistical features in complex data sets, yet computers lack the robustness and generalizability associated with human learning. Specifically, humans can leverage previously learned knowledge to avoid needing large training sets, transfer knowledge from one task to other related tasks or contexts to accelerate learning, continuously update their learned knowledge, and adapt to time-varying contexts and environments. We are interested in brain-inspired or biomimetic algorithms that advance the state of the art and have the potential for revolutionary progress on these or related research challenges.
  • Brain-inspired Computing. Recent examples of commercially-developed neuromorphic hardware have focused on power consumption and density advantages over conventional hardware, but brain-inspired circuits also hold promise for novel computing architectures that can address problems that traditional von Neumann architectures cannot (or at least not within polynomial time or better). Research proposals on brain-inspired computational capabilities should describe how the research will fill gaps in computing or how it will enable novel artificial intelligence algorithms.

Models of brain structures and functions should be biologically plausible and demonstrate consistency with the current understanding of the structure and function of the brain components being modeled. Brain-inspired algorithms or hardware need not remain entirely faithful to the biological details where other approaches are more reasonable, but neuroscience evidence should be used for benchmarking and theoretical guidance. Proposals should provide a definition of success based on specific metrics and/or one or more challenge problems.

Prospective proposers are highly encouraged to contact the Program Officer prior to developing a full proposal to discuss the proposed research and how it will advance the state of the art, the approximate cost for a three (3) to five (5) year effort, and any specific submission target dates.

DR. HAL S. GREENWALD, AFOSR/RTA2
Email: neuroscience@us.af.mil

 
USAF/USSF logos