United States

Artificial Intelligence (AI) has become integral to critical applications, yet the lack of robust mechanisms for safety and explainability poses significant security concerns. To address this, ObjectSecurity LLC has developed a novel AI/ML trust analysis tool—funded through SBIR and internal R&amp;D—that detects vulnerabilities in AI systems, enhances explainability, and enables targeted patching of identified weaknesses within an AI model.

This poster presents the technical mechanisms behind our approach, demonstrating how it detects and mitigates vulnerabilities in AI/ML models. The tool conducts a layer-by-layer analysis to identify key risks, including susceptibility to adversarial attacks and inefficiencies caused by overfitting. It improves model transparency by uncovering hidden relationships between tokens and optimizing neuron utilization, enabling targeted fine-tuning for maximum efficiency. By generating actionable metrics and applying custom LoRA masks, the tool facilitates rapid model adjustments, enhancing robustness, accuracy, and adaptability.

By fostering trust, confidence, and accountability, our solution supports the deployment of AI systems in high-stakes environments.

NAML 2025

Vulnerability Analysis of AI/ML Systems

ai remediation

secure ai

trustworthy ai

### Welcome to the 9th 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. 
<br>Naval Information Warfare Center Pacific is partnering with AFCEA San Diego chapter to host the workshop.

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>

Please log in to explore this event. You need to log in with the email address you registered with. Access credentials have been sent to your email. 

Please be sure to check your spam and other email folders if you do not see an email confirmation right away.

Please log in to explore this event.

Registration for this workshop has now closed. We appreciate your interest and hope to welcome you at a future event. If you’d like to hear about similar upcoming events, please check our website or join our mailing list.

Registration for this workshop has now closed.

The 9th annual workshop on Naval Applications of Machine Learning 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. 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 and to build and strengthen collaborations within the defense research community.

Artificial Intelligence (AI) has become integral to critical applications, yet the lack of robust mechanisms for safety and explainability poses significant security concerns. To address this, ObjectSecurity LLC has developed a novel AI/ML trust analysis tool—funded through SBIR and internal R&D—that detects vulnerabilities in AI systems, enhances explainability, and enables targeted patching of identified weaknesses within an AI model.

This poster presents the technical mechanisms behind our approach, demonstrating how it detects and mitigates vulnerabilities in AI/ML models. The tool conducts a layer-by-layer analysis to identify key risks, including susceptibility to adversarial attacks and inefficiencies caused by overfitting. It improves model transparency by uncovering hidden relationships between tokens and optimizing neuron utilization, enabling targeted fine-tuning for maximum efficiency. By generating actionable metrics and applying custom LoRA masks, the tool facilitates rapid model adjustments, enhancing robustness, accuracy, and adaptability.

By fostering trust, confidence, and accountability, our solution supports the deployment of AI systems in high-stakes environments.

poster

Throughout the Fleet, operators are faced with complex tasks that require collaborative interactions between heterogeneous teams and software systems. This is especially embodied by Naval planners working through the Naval Planning Process (NPP). Although recent work has focused on modernizing the way this process unfolds through the introduction of digital systems like the Advanced Multi-echelon Planning Tool (AMPT), we believe that empowering planners with more natural ways to interact with tools like this will result in better user experiences and more streamlined workflows to bring them from initial concept to the final planning artifact. Achieving this requires the concurrent development of machine learning (ML) enabled processes such as automated speech recognition (ASR), user intent recognition, information extraction and synthesis, and in-context learning. Ultimately our goal brings all these pieces together such that Naval planners can employ the unstructured domain specific language they use with one another to directly interface with the planning systems needed to complete their jobs.

Natural Language Interfaces for Naval Planners

Mnemosyne is an advanced Retrieval Augmented Generation (RAG) system inspired by the Greek goddess of memory, who learned something new every day and never forgot, ultimately becoming the most knowledgeable of all the Titans. This system embodies continual learning and knowledge accumulation, forming the foundation of its architecture.
Mnemosyne employs a Mixture of Pipelines (MOP) approach, derived from mixture-of-models architectures, to address diverse question types. Specialized pipelines handle general knowledge queries, quantitative (counting) questions, temporal analyses (summarizing time-based data and trends), and comprehensive summaries (synthesizing entire corpora).
To enhance performance, Mnemosyne integrates several intelligent agents. The RAG Fusion Agent generates alternative perspective questions that enrich responses. The Pipeline Routing Agent directs queries to the appropriate pipeline, while the Rank Shuffling Agent increases result diversity by introducing random rank changes. A Validation Agent ensures response accuracy by cross-checking summaries with source documents using alternative models.
Advanced prompt engineering techniques from the latest literature further refine the system’s capabilities. Chain of Thought reasoning improves response accuracy by allowing intermediate steps, Agent-Selected kNN Few-Shot Learning selects relevant examples from curated libraries, and answer-choice shuffled ensembling improves robustness through aggregated outputs. Mnemosyne also employs Reciprocal Rank Fusion to combine and rank the search matches of multiple pipelines, questions, and rank shuffles, promoting high-quality responses.
A key feature is Mnemosyne’s modularity and flexibility, allowing interaction with multiple models and databases to accommodate varying compute and storage capacities. This flexibility allows the use of state-of-the-art models or deployment of lightweight models for edge applications.
An optional knowledge graph integration enhances functionality by using a graph as the vector store. This enables JSON-based proximity representations to visualize relationships around matched entities and graph-based ranking refinements to improve entity matches. Mnemosyne can query graph databases like Neo4j, generating Cypher queries to extract and analyze relevant relationships.
Mnemosyne represents a significant advancement in enterprise-grade RAG systems by integrating multiple agents, innovative prompt engineering, and optional knowledge graph capabilities. Its adaptable architecture transforms data into actionable insights while continuously refining knowledge, embodying the memory and wisdom of its mythological namesake.


Mnemosyne: Architecting an Enterprise-Grade Retrieval Augmented Generation Pipeline with Mixture of Pipelines

Underwater environments present a significant challenge for imaging systems due to the dynamic interplay of light attenuation, scattering, and absorption. These factors severely degrade visibility and signal quality, complicating critical tasks such as object detection, recognition, and tracking. The turbid nature of underwater scenes further exacerbates these challenges, as light propagation along the line-of-sight (LOS) between the transmitter and detector is highly susceptible to environmental perturbations. These limitations are particularly critical in naval applications, where robust imaging systems are essential for operations such as surveillance, navigation, and search-and-rescue missions.
To address these challenges, we have developed a novel imaging system that combines multipolarized single-frequency illumination and single-photon detection. Our experimental setup includes lasers transmitting 532nm light in four distinct polarization states. To process the photon returns, we designed and engineered a meta-surface nanostructure attached to our imaging setup. This nanostructure spatially splits incoming photons into four distinct polarization channels captured by our single photon detectors, producing a total of 16 unique polarization-resolved images per capture. This modality allows us to systematically analyze how each polarization state interacts with underwater environments characterized by varying levels of turbidity.
We show that different polarization states can accentuate specific target features while mitigating the effects of scattering and absorption. By isolating and leveraging the most effective polarization states, we enhance the quality of acquired images, enabling downstream tasks such as object detection and recognition to operate more effectively. This approach not only improves image clarity but also provides a foundation for optimizing imaging systems under diverse environmental conditions.
To further refine our understanding, we employ information-theoretic approaches to quantify the information content provided by each polarization state. This analysis helps discern the unique contributions of each channel, offering deeper insights into the interaction between light and underwater environments.
The integration of multipolarized imaging and single-photon detection represents a scalable framework for adaptive underwater imaging systems. These systems have significant implications for naval applications, offering enhanced object detection and recognition capabilities that can improve situational awareness in operations ranging from mine detection to underwater robotics. By addressing the limitations of traditional imaging methods, this work paves the way for advanced optical systems capable of operating robustly in the most challenging underwater conditions.


Multi-polarization Underwater Object Recognition with Metasurfaces-based Lidar Imagery

One of the problems with NN-based object detectors is a problem endemic to deep learning-based object detection models in general. Deep learning puts an enormous amount of trust in the training set to represent all configurations detectable objects within the imagery to be captured will take. Our contention here is that while NN-based object detectors’ ability to find objects is extremely valuable, NN-based object detectors tends to be “brittle” in the sense that when an object gets obscured by some means or changes its configuration beyond the scope of the confines of the training set, the NN-based object detectors-generated object detection bounding box disappears; thus, generating a false negative in current and (possibly) subsequent video frames. We assert that this phenomenon can be overcome by means of inserting an “adjudicator” based on elastic shape analysis (ESA). Our adjudicator algorithm, which we name LISA or Light Informed Shape Analysis operates by capturing and scoring the shape of the object contained in a prospective region of interest using only the present illumination on the object. This reliance on nothing more than current illumination in our scene has the effect of dismissing the need for a large and comprehensive training set. LISA creates an object detection score of the object’s illumination-defined shape in a prospective ROI by means of aggregating the detected object’s illumination-defined shapes that neural network-based object detectors previously captured. This aggregation occurs by using a weighted Karcher mean, and the scoring occurs by using the Path Straightening Algorithm (PSA). Combining LISA with YOLO leads to SOLO or Shapes Only Look Once and ultimately serves to enhance YOLO’s inference step. We provide evidence of SOLO’s false-negative mitigating ability for YOLO by considering a series of examples depicting objects that are partially concealed and operating in challenging video acquisition circumstances.

Mitigating False Negatives in Neural Network Object Detectors through Light Informed Shape Analysis (LISA)

The presentation will discuss a notional framework for the application of generative AI in addressing human machine teaming being used in guiding ONR investments in the Science of AI Applied Research Portfolio. The extension of the framework to address a challenge problem posed by RADM Rothenhaus (Chief of Naval Research - CNR) to with decisin making of Commanders of Carrier Strike Grouos (CCSGs) is introduced, as well as a proposed FY27 Innovative Naval Prototype (INP) program that responds to the CNR challenge. Decision Driven Command and Control (D2C2) is introduced with an invitation to discuss Science and Technology problems implied by the INP.

Generative AI for Human Machine Teaming (HMT) Applications

In this talk we will introduce the notion of topological nodes in neural networks, focusing specifically on the initial applications of a category called circular nodes. Circular nodes can be implemented as node pairs with shared constraints allowing one to encode angular information—like signal phase—within the network. We will present a brief discussion of the theory behind these nodes and present a novel implementation of these compatible with existing deep learning frameworks. Using this new implementation, we recreate examples from the original paper and apply circular nodes to toy classification tasks. Finally, we will demonstrate initial success capturing phase changes in radio frequency signals and conclude with a discussion of future use cases.

Topological Nodes and their Applications

Panel Discussion: MAPT drivers: symbiotic relationships between the defense and commercial sectors

Background: This study explores factors influencing attrition in the U.S. Marine Corps by analyzing participant data from the Recruit Assessment Program (RAP) and employing machine learning techniques. 
Methods: The primary focus is to predict contract fulfillment, defined as attrition, fulfillment, or retention, and secondly, to identify predictors of different types of discharge. The study population included 53,156 male recruits at Marine Corps Recruit Depot, San Diego, who enlisted between 2013 and 2018, with a five-year follow-up period. 
Results: The analysis revealed that 15.0% of recruits did not meet their contractual obligations, 58.6% fulfilled them, and 26.4% exceeded them. Machine learning models identified in-service factors—such as promotion status, military occupation, and deployment rate—as the strongest predictors of both attrition and retention, whereas pre-service mental and behavioral health factors had minimal influence. Promotion status, as reflected by grade expectations, emerged as the most important predictor, with failure to meet early career promotions increasing the risk of attrition. In contrast, pre-service mental health variables, including depression, PTSD, ACE, and problematic anger, were not predictive of career outcomes. 
Conclusions: These findings suggest that in-service experiences and career progression play a more substantial role in determining long-term success in the military than pre-service mental health factors. These results highlight the potential for improving retention through enhanced support services during early military service rather than relying primarily on pre-service health screening.

Comparing Pre-service and In-service Predictive Factors that Lead to Attrition Among Active Duty Marine Recruits

Delivered by Matt Beekman, AFCEA San Diego Chapter President and Katie Rainey, director of Science and Technology, Intelligence, Surveillance & Reconnaissance Department, Naval Information Warfare Center Pacific and NAML 2025 General Chair.

Opening Remarks

Power in the modern world - military, economic, geopolitical - is built on a foundation of computer chips. America has maintained its lead as a superpower because it has dominated advances in computer chips and all the technology that chips have enabled. (Virtually everything runs on chips: cars, phones, the stock market, even the electric grid.) Now that edge is in danger of slipping, undermined by the naive assumption that globalising the chip industry and letting players in Taiwan, Korea and Europe take over manufacturing serves America`s interests. Currently, as Chip War reveals, China, which spends more on chips than any other product, is pouring billions into a chip-building Manhattan Project to catch up to the US.

In Chip War economic historian Chris Miller recounts the fascinating sequence of events that led to the United States perfecting chip design, and how faster chips helped defeat the Soviet Union (by rendering the Russians’ arsenal of precision-guided weapons obsolete). The battle to control this industry will shape our future. China spends more money importing chips than buying oil, and they are China`s greatest external vulnerability as they are fundamentally reliant on foreign chips. But with 37 per cent of the global supply of chips being made in Taiwan, within easy range of Chinese missiles, the West's fear is that a solution may be close at hand. 

Premium content

Downloads

Next from NAML 2025

Natural Language Interfaces for Naval Planners

Similar lecture

Mission-Critical AI: Automating Trust, Security, and Performance at Scale

Stay up to date with the latest Underline news!

PRESENTATIONS

CONFERENCES

COMPANY

RESOURCES