United States

AI systems deployed in military contexts present unique risk profiles not addressed by traditional cybersecurity models. This work introduces a hybrid threat modeling and risk validation methodology that aligns AI-specific attack vectors from MITRE ATLAS with two risk control frameworks: the OWASP Risk Rating Methodology (RRM) and the US Navy&#39;s Operational Risk Management (ORM) model. Our approach begins with threat enumeration using MITRE ATLAS to identify attack paths relevant to military AI systems, including prompt injection, data poisoning, and misuse of model outputs. We then apply OWASP RRM to score technical likelihood and impact. These results are translated into ORM terms of Severity, Probability, and a Risk Assessment Code (RAC) to support risk decisions by command elements. We validate this method through structured analysis of Retrieval-Augmented Generation (RAG) systems used for mission planning and decision support. Failure modes such as hallucinated intelligence summaries and source contamination are modeled and scored, producing risk matrices in both OWASP and ORM formats. This enables both cybersecurity teams and operational leadership to act on a shared, mission-aligned view of AI risk. The resulting framework provides a practical, adaptable methodology for AI risk assessment in coalition environments.

NAML 2026

A Hybrid Threat Modeling and Risk Validation Framework for Mission-Critical AI Systems

resilient ai

adversarial ai

ai vulnerability assessment

mitre atlas

owasp

navy orm

ai risks

ai security

robust ai

### Welcome to the 10th annual workshop on Naval Applications of Machine Learning!

NAML is the premier annual event showcasing current machine learning research relevant to Naval applications. We welcome participation from the defense community, industry, and academia. The NAML workshop is an opportunity for researchers from Navy research and development centers and our partner organizations to exchange technical ideas and to come together as a team. 
Naval Information Warfare Center Pacific is partnering with AFCEA San Diego chapter to host the workshop.
#### On-Demand Access

* This is an on-demand site and all the recorded content from the workshop is available here. 
* Access to the recordings is exclusively available to registered participants.
* Each registrant has received an email notification, with instructions on how to access the content.

[![](https://assets.underline.io/markdown_image/1/image/60cd12aa376c191d3467ea2ee60944ea.png)](https://naml2026.org/)

The goals of NAML are:

* To spread awareness of current machine learning research relevant to Naval applications
* To connect machine learning researchers with experts in military requirements
* To build and strengthen collaborations within the defense research community

#### *"What we do as a Navy is a human business. Don’t pass up an opportunity to come together as a team. Don’t pass up leadership opportunities, and don’t outsource them to a PowerPoint. If you have an opportunity to come together as a team, to increase belonging, do it."*
Admiral John M. Richardson, 31st Chief of Naval Operations
<br><br>

This is an on-demand site. All recorded content from NAML 2026 is available here and access to it is exclusively available to registered participants.

**Please log in using the email address you registered with.** If you have any issues logging in, please contact us at <naml2026@underline.io>.

Let’s celebrate the past decade by investing in the next. 🚀


Please log in to explore this event.

This is an on-demand site. All recorded content from NAML 2026 is available here and access to it is exclusively available to registered participants.

**Explore the latest advancements in machine learning tailored to Naval applications 🔎**

With on-demand access to NAML 2026, you can watch all lectures & browse poster presentations at your convenience, while also engaging directly with presenters via the Underline platform - ask questions, exchange ideas, and stay connected.

**If you still haven’t registered, you can do so now.**
👇
**Take advantage of a special one-time offer: full access for just $99, with complimentary registration for students, government, and military employees.**

**Get access here: https://naml2026-ondemand.paperform.co/**

Please check your spam folder, or contact our support team at **<naml2026@underline.io>** if you registered, but still haven't received an access email.

Let’s celebrate the past decade by investing in the next. 🚀

Please register!

The 10th Annual Workshop on Naval Applications of Machine Learning (NAML 2026) was held in San Diego, California, March 2-5, 2026. NAML is a forum for interdisciplinary experts from government, industry, and academia to discuss results of ongoing research on theory and applications of machine learning to address some of the most challenging problems faced by our naval forces. 



AI systems deployed in military contexts present unique risk profiles not addressed by traditional cybersecurity models. This work introduces a hybrid threat modeling and risk validation methodology that aligns AI-specific attack vectors from MITRE ATLAS with two risk control frameworks: the OWASP Risk Rating Methodology (RRM) and the US Navy's Operational Risk Management (ORM) model. Our approach begins with threat enumeration using MITRE ATLAS to identify attack paths relevant to military AI systems, including prompt injection, data poisoning, and misuse of model outputs. We then apply OWASP RRM to score technical likelihood and impact. These results are translated into ORM terms of Severity, Probability, and a Risk Assessment Code (RAC) to support risk decisions by command elements. We validate this method through structured analysis of Retrieval-Augmented Generation (RAG) systems used for mission planning and decision support. Failure modes such as hallucinated intelligence summaries and source contamination are modeled and scored, producing risk matrices in both OWASP and ORM formats. This enables both cybersecurity teams and operational leadership to act on a shared, mission-aligned view of AI risk. The resulting framework provides a practical, adaptable methodology for AI risk assessment in coalition environments.

technical paper

Sea Strike 2043 sketched a “command center of the future” but left the core question open: what brain actually drives it? Atombeam’s Persistent Cognitive Machine (PCM) is a working prototype of that brain, implemented as Kennedy PCM for maritime command and control. Instead of treating sensor and intelligence feeds as logs or text tokens, PCM encodes them as trajectories on a continuously evolving geometric manifold, providing true persistent cognition at theater scale.

We will demonstrate Kennedy PCM on a notional island-defense vignette inspired by Sea Strike: a Red amphibious invasion attempting to force a strait against Blue forces. The system ingests multi-domain, multi-timescale data (radar, AIS, UxS swarms, ISR, cyber events, operator inputs) and (1) builds a layered manifold of the battlespace, (2) discovers and retains operational patterns over months of simulated conflict, and (3) searches this learned geometry for courses of action under explicit constraints (ROE, risk, asset limits). In the demo, Kennedy uses a year of prior engagements to predict Red’s feint and strait-transit timing, identifies a geometry-driven COA that exploits an emergent 8-second vulnerability window on the landing force, and explains its recommendations via “geometric certificates” that can be replayed and audited by watch teams.

We explicitly contrast this with today’s LLM/agent stacks. LLMs must repeatedly “start from zero,” re-discover context every shift, and rely on brittle toolchains and cloud-scale compute; they do not naturally remember years of operations or share that memory across units. PCM instead learns continuously, storing experience directly in geometry, with a footprint small enough to operate within a single system-on-chip. This makes it practical to embed Kennedy-class cognition in thousands of UAS and other edge platforms, allowing them to operate coherently and complete missions even under EMCON or total loss of reach-back communications. Multiple PCMs federate by exchanging compact geometric patterns rather than raw data or prompts, enabling learning across strike groups while respecting bandwidth limits and classification boundaries.

We will conclude with implications for NIWC Pacific’s cognitive C2 vision, JADC2 integration, human–machine teaming experiments, and scaling PCM to live Navy data and future CCA/NGAD-style, UAS-heavy kill webs. Preferred format is a 20-minute short talk with live demo; the material can be adapted to a lightning talk if required.



Kennedy PCM: Persistent Geometric Cognition for the Naval
Command Center of the Future

If two experts disagree on a test, we may conclude both cannot be 100% correct. But if they completely agree, no possible evaluation can be excluded. This asymmetry in the utility of agreements versus disagreements is explored here by formalizing a logic of unsupervised evaluation for classifiers. Its core problem is computing the set of group evaluations that are logically consistent with how we observe them agreeing and disagreeing in their decisions. Statistical summaries of their aligned decisions are inputs into a Linear Programming problem in the integer space of possible correct or incorrect responses given true labels. Obvious logical constraints, such as, “The number of correct responses cannot exceed the number of observed responses,” are inequalities. But in addition, there are “axioms”—universally applicable linear equalities that apply to all finite tests. The practical and immediate utility of this approach to unsupervised evaluation using only logical consistency is demonstrated by building no-knowledge alarms that can detect when one or more LLMs-as-Judges are violating a minimum grading threshold specified by the user.

Logical Consistency Between Disagreeing Experts and Its Role In AI Safety

Fuel distribution and logistics is a complicated but necessary field. It requires a greater understanding of the environment than just getting from A to B, it needs everything in between. That is precisely what Logistics Bulk Fuel Expeditionary Analysis and Simulation Tool (LOGBEAST) brings to the table for the United States Marine Corps (USMC). 

Currently, Marines must engage in laborious manual planning using decentralized data in spreadsheets—a process that is time-consuming, tedious, and prone to human error. Additionally, it is not accessible by the wider Marine Corps or Department of War’s (DoW) Logistics Enterprise. Using classic optimization modeling techniques along with data acquired from throughout the Indo-Pacific theater, LOGBEAST takes in a detailed configuration from the user and returns a comprehensive fuel delivery plan with estimated delivery times that take into account: physical transfer rates of fuel, queuing of trucks, slow-downs due to congestion in population centers, and other factors impacting ground travel with large vehicles.

Currently deployed in DoW’s ADVANA platform, LOGBEAST provides a centralized industrial grade optimization model built for speed and efficiency using software that determines the optimal schedule of fuel deliveries from the hub to downstream squadrons positioned at spokes, then generates the shortest paths along real roads, routing around physical restrictions caused by civil infrastructure and/or user identified restricted zones.

LOGBEAST accounts for maximum realism such as the visualization of historical weather trends layered on high resolution satellite imagery, and user selection of route optimization by either time or distance and/or obstacle avoidance, plus provides an analysis of potential locations for fuel equipment laydown placement at certain locations.

Currently, LOGBEAST leverages a Retrieval-Augmented Generation (RAG) Large Language Model (LLM) enriched with USMC fuel planning doctrine enabling context-aware and data grounded responses to quickly and precisely answer questions for users. Future AI plans include the modeling fuel distribution via sea and air. Given LOGBEAST’s focus on the Indo-Pacific Theater, incorporation of the naval layer for island-to-island fuel transfers and aerial transport is critical to model the full scope of the problem USMC faces in the Pacific. To advance efforts, an agentic AI approach may be pursued that would allow the LLM to select not only the best models/data needed for the mission but to proceed with utilizing said tools and provide the user with the generated outputs. 

Ultimately, LOGBEAST fills a void felt by the Marine Corps that would otherwise render the fuel planning process vulnerable to slowdowns, human errors, broad generalizations, and a lack of centralized data sharing amongst all Services.


Logistics Bulk Fuel Expeditionary Analysis and Simulation Tool (LOGBEAST)

Presenting PEO C4I as an acquisition command, how S&T is done at PEO C4I and how industry/academia/government can help with AL/ML.

PEO C4I Science and Technology (S&T) at NAML

The motion response of a surface vessel in following or stern quartering seas may include a sudden uncontrollable turn, known as broaching. The physical reason for broaching is an instability in yaw caused by a combination of the hull forces, thus the rudder action may not have an effect at all. Broaching-to is not a new problem; it has been known since the Age of Sail [1], so experienced operators are aware of it.

A commonly known precursor to the broaching phenomenon is surf-riding. During a surf-ride, a ship accelerates or decelerates to match the wave celerity such that the ship travels with the incident wave profile. In the realistic sea, these periods of matching the speed of the wave profiles appear and disappear with acceleration, making analysis complex. Given the common frequencies inherent to the sea and typical ship speeds, surf-rides are typically linked with periods of “high-run” [2]. It follows that the instantaneous speed of the ship as well as the speed of the wave profile at the ship’s center of gravity are important parameters that inform surf-riding and potentially broaching.

The proposed method introduces a machine learning framework that links time series realizations of ship speed and wave celerity to the instantaneous yaw of the ship. A Long-Short Term Memory (LSTM) [3] neural network is trained with various combinations of tri-chromatic wave time series and surge velocity time series simulated with the reduced-order SimpleCode [3] as input and with yaw time series generated by SimpleCode as targets. The results of the trained neural network are evaluated against SimpleCode in ability to estimate large yaw events.

References

[1] Spyrou, K. J. “Perceptions of Broaching-To: Discovering the Past”, Chapter 22 of Contemporary Ideas on Ship Stability, Neves, M.A.S., Belenky, V., de Kat, J. O., Spyrou, K. and N. Umeda, eds., Springer, ISBN 978-94-007-1481-6, pp. 399-414, 2011.

[2] Themelis, N., Spyrou, K., and Belenky, V. “Surf-riding in Multi-chromatic Seas: “High-runs” and the Role of Instantaneous Celerity,” in Ocean Engineering, Vol. 120, pp 230-237, 2016.

[3] Hochreiter, S., and Schmidhuber, J., “Long Short-Term Memory,” Neural Computation, Vol. 9, No. 11, pp. 1735–1780, 1997.

[4] Weems, K. and Belenky, V. “Extended Fast Ship Motion Simulations for Stability Failures in Irregular Seas,” Proceedings of the 13th International Conference on Stability of Ships and Ocean Vehicles, Kobe, Japan, 2018.

Ship Broaching Prediction with Machine Learning

Reliable sonar interpretation is essential for many naval applications, from autonomous undersea vehicles to undersea surveillance and mine detection. However, deep learning models that achieve high performance in this domain often rely on large, resource-intensive architectures that are impractical for size, weight, and power (SWaP) constrained autonomous and edge-deployed buoy systems. Knowledge distillation (KD) provides an efficient alternative by transferring useful knowledge from large teacher networks to lightweight student models, allowing smaller models to perform the same task with reduced computational cost. While KD has shown promise across various audio tasks, its limited exploration in sonar arises from the challenge that existing methods often neglect the fine-grained texture cues critical for reliable detection and classification in complex, undersea acoustic environments. To address this gap, we leverage the Structural and Statistical Audio Texture Knowledge Distillation (SSATKD) framework, designed to enhance sonar perception across both passive and active modalities. SSATKD fuses high-level contextual features with low-level texture cues using two complementary modules: an edge detection module that identifies structural textures reflecting periodic patterns (i.e., repetitions) present in both spectrograms and sonar imagery (e.g., side-scan), and a statistical texture knowledge extractor that characterizes signal variability and distributional differences across environments. In passive and active sonar, SSATKD improves vessel classification and target segmentation respectively by leveraging structural and statistical texture cues that help improve with target recognition and precise localization. Across real-world sonar datasets, SSATKD extends prior work by providing stronger empirical validation in operational domains and improving texture modeling through a redesigned objective function, resulting in significant performance gains while reducing model parameters and computational cost for deployment on embedded, SWaP-constrained naval platforms. These results highlight the potential of texture-aware knowledge distillation to bridge perception tasks across sonar modalities and advance autonomous sensing capabilities for undersea operations.

LINE PROGRAM CUSTOM MARKING (LIMITED-RELEASE ONLY) - STANDARD MARKINGS ARE NOT APPLICABLE. USE THE FOLLOWING INSTEAD: Distribution authorized to U.S. Government agencies and their contractors. Other requests for this document must be referred to the MIT Lincoln Laboratory Technology Office. This material is based upon work supported by the Under Secretary of War for Research and Engineering under Air Force Contract No. FA8702-15-D-0001 or FA8702-25-D-B002. Any opinions, findings, conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the Under Secretary of War for Research and Engineering.

Texture-Aware Knowledge Distillation for Efficient Passive and Active Sonar Analysis

Bottlenose dolphins produce various whistle types, including signature whistles, shared whistles, copies of conspecifics’ signature whistles, and variant whistles. While signature whistles have been well studied, due to their individual distinctiveness, much less has been done to identify individuals from non-signature whistle types. This study trained m18 convolutional neural networks (CNNs) to classify dolphin whistle producers from the 1D time-series inputs of whistle recordings of 20 dolphins at the U.S. Navy Marine Mammal Program. The model trained exclusively on signature whistles achieved 93.7% accuracy (random chance = 5%). The same model trained using all whistle types achieved 85.2% accuracy overall and performed slightly better on signature whistles (96.1%), suggesting the broader repertoire and larger dataset improved classification. Shared whistles were classified with 92.5% accuracy and variant whistles still were classified at 72.1%, suggesting that some individually identifying information is encoded across these types. Signature whistle copies proved difficult to classify (17.2%). With localization remaining a challenge in social groups, this study offers a promising step toward caller identification in well-studied populations. These results suggest that deep learning applied to raw audio time series data can learn individual whistle repertoires, offering new opportunities for acoustic monitoring of odontocetes.

Identifying Dolphin Whistle Producers with Deep Learning: Moving Beyond Signature Whistles

Inverse problems in deep learning that infer initial parameters from a final output are challenging, particularly when the generative process is a "black-box" function. These problems are often ill-posed and non-bijective, meaning a single output can map to multiple valid inputs. 

We introduce a novel framework that employs a conditional denoising diffusion model to predict the multimodal distribution of rendering parameters from a 2D image. We demonstrate our methodology on a practical use case: inferring camera and lighting parameters for a rendered image of a 3D vehicle model within a complex, non-differentiable Blender pipeline. Our approach trains a conditional diffusion model on a dataset of image-parameter pairs generated by the black-box. 

The architecture uses a pre-trained ResNet encoder to extract image features and a custom 1D U-Net with cross-attention to predict the noise in a corrupted parameter vector. By conditioning the denoising process on image features, the model learns the non-linear mapping between visual cues (e.g., shadows, highlights, and perspective) and the corresponding rendering parameters. 

We demonstrate how our trained model can generate a diverse, accurate distribution of plausible parameter sets from a single rendered image. By sampling the reverse diffusion process multiple times, the model captures the solution space's inherent multimodality. Quantitative analysis, measured by the Mean Squared Error between the ground truth and the best-predicted mode, illustrates our approach's high fidelity. Future work will focus on reducing training queries to the computationally expensive black-box, enhancing efficiency in applications like scientific simulation and procedural content generation.

Inverse Black Box Diffusion Modeling

Space-based Electro-Optical (EO) imaging sensors continued to be launched at an increasing rate. These systems may be used to help detect illicit maritime activity. However, challenges exist in the use of deep learning-based computer vision algorithms to derive maximum value from this growing data source. In particular, maritime ship detection and classification algorithms have demonstrated performance loss in the presence of novel imagery classes or out-of-distribution images. One means for mitigating this issue is the incorporation of Automated Identification System (AIS) vessel track data. Relatively little work has explored the use of multi-modal computer vision models to make use of satellite imagery and time coincident AIS data for image classification. This paper develops novel multi-modal imagery and AIS models to explore whether the addition of AIS data can improve model robustness. Image pixels are dropped to zero at varying rates to emulate varying degrees of out-of-distribution data in order measure robustness. It is shown that the multi-modal imagery and AIS models can yield an up to a 13.3% increase in the average precision to identify a ship in imagery when 67% of pixels have been dropped from an image. In all models examined, the multi-modal models outperformed the imagery-only models across the vast majority of cases examined. This confirms that the additional AIS modality can improve model robustness.

Characterizing the robustness of multi-modal AIS and imagery models for vessel image classification

The rapid evolution of Machine Learning (ML) and Artificial Intelligence (AI) particularly Large Language Models (LLMs) and agentic technologies, presents exceptional opportunities and new complexities for Navy and Marine Corps mission planning and execution. For leaders, planners, and system owners, the path to successful, scalable AI adoption starts with a clear understanding of the foundational elements that make enterprise AI truly impactful. 
This session will explore critical infrastructure considerations, from dataset size evaluations and interpreting tokenization metrics to anticipate GPU resource needs, to mapping storage solutions for high-performance AI model training and inference. We’ll cover the nuances of vector storage requirements, common “gotchas” encountered when upscaling from pilot to production, and the operational benefits of multi-cloud and hybrid architectures in delivering both resilience and flexibility across diverse DoN environments. 
Illustrative scenarios will showcase how agentic AI workflows and LLMs can enhance everything from intelligence analysis to logistics and communications, demonstrating how specific environmental and architectural choices directly influence operational outcomes. 
Examples of these applied AI use cases include:
• Course of Action development and decision support
• Battlefield awareness and mission readiness 
• ISR and geospatial analysis
• predictive maintenance 
• recruiting retention, promotion, attrition, etc.

Attendees will also receive insights on optimizing data architectures and compute environments to support continuous innovation, agility, and mission alignment. 
Special consideration will be briefly given to AI at the edge, highlighting how edge-based solutions that integrate with cloud can support secure, high-performance AI even in disconnected and austere environments. 
By participating in this session, DoN stakeholders will leave with the tools and frameworks needed to systematically define requirements and prepare their environments, enabling the successful integration and scaling of AI-powered capabilities for sustained mission advantage.


Premium content

Downloads

Next from NAML 2026

Kennedy PCM: Persistent Geometric Cognition for the Naval Command Center of the Future

Similar lecture

Operationalizing AI Risk Tolerance: Framework-Agnostic Taxonomy, Calibration, and Composite Risk Scoring

Stay up to date with the latest Underline news!

PRESENTATIONS

CONFERENCES

COMPANY

RESOURCES