United States

As with any tool, there are risks to misusing Large Language Models (LLMs). However, unlike with standard tools, even correct use of LLMs can cause harm – consider hallucinations: unfounded replies that are simply made up. 

Technological changes, such as upgrading a model to a newer, more performant version are relatively easy. Changing human behavior – how people use these tools – is much harder: established usage patterns can be hard to budge. 

To address this issue, our team created an LLM demo website for a client founded on human design principles to prioritize responsible usage. We emphasize the following features:
• Provide guidance on how models can be biased and can hallucinate 
• Provide information on which LLM was chosen and why
• Explain how the model was trained
• Reveal what information is saved, and what is not
• Calculate and provide automated diagnostics of model performance in real-time 
• Empower users to upload their own documents to see how well the model works

We believe this is a differentiated approach that focuses on the less malleable, but more important aspect of LLM interaction: the user. We believe this approach could and should be more widely used throughout the U.S. Government, and particularly within the Department of Defense (DoD). 


NAML 2025

Human Centric Design to Foster Responsible LLM Use

retrieval-augmented generation (rag)

large language model (llm)

human factors

chatbot

It is my great pleasure and honor to welcome you to the 64th Annual Meeting of the Association for Computational Linguistics (ACL 2026), held in San Diego, California, USA, July 2–7, 2026. ACL 2026 continues our fields tradition of excellence in scholarship, innovation, and inclusivity, and I am deeply grateful to the many volunteers who have worked tirelessly to bring this event to life.

It is exciting to see the many developments in the field reflected at our conference, such as the advancement of large language models (born out of our community) into many application directions. This is not without challenges, partly due to unsolved problems, and partly due to misuse of the technology. All this is reflected well at ACL 2026. The conference also reflects that the field is not only about advancing technology, but also the science of fundamentally understanding language and how it is use by many different speaker communities. 
[Full Message from General Chair Philipp Koehn](https://drive.google.com/file/d/1QiJFOmMjvDgsDo3i1XzlQnImxe3RU_Cy/view?usp=sharing) 
<html><button style="display: inline-flex; align-items: center; justify-content: center; white-space: nowrap; border-radius: 9999px; font-weight: bold; background: #7c3aed; color: white; font-family: 'Space Grotesk', sans-serif; height: 40px; font-size: 16px; padding: 0 20px; border: none; cursor: pointer" onclick="window.open('https://drive.google.com/file/d/1QiJFOmMjvDgsDo3i1XzlQnImxe3RU_Cy/view?usp=sharing')">Conference Handbook</button></html>

To access this event page, you need to log in with the **email address you registered with**. Access credentials will be sent to your email from Underline - subject line "Welcome to ACL 2026". Please be sure to check your spam email folder if you do not see an email confirmation right away.

To access this event page, you are required to register.
Please complete your registration to continue.

We recommend reading [**the registration information**](https://2026.aclweb.org/registration/) first.

**Online Registration Form**: https://acl.swoogo.com/acl2026

Registration Required

ACL 2026

The 64th Annual Meeting of the Association for Computational Linguistics (ACL 2026) will take place in San Diego, California, from July 2 - 7, 2026.

poster

### 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. 
 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

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.

The objective of this research was to evaluate machine learning-based optical character recognition (OCR) to automate recognition of aircraft side numbers on the carrier flight deck. In this research, we developed a software pipeline that utilizes a pan-tilt-zoom (PTZ) camera to achieve this automation. The pipeline detects all aircraft in view, a PTZ camera to obtain high resolution images of each aircraft, then OCR to extract the side number text. The final algorithm is demonstrated using a combination of stationary aircraft.


Automated Aircraft Sidenumber Recognition using Computer Vision for Naval Aircraft Logistics

Spiking Neural Networks (SNNs) [1], which are inspired by the functioning of the brain, offer a promising approach to artificial intelligence with significant advantages in energy efficiency. This makes SNNs particularly suitable for use in underwater edge devices, such as unmanned underwater vehicles (UUVs), which are often constrained by limited battery capacity. However, current SNN-based methods for underwater acoustic data representation learning remain relatively underexplored. In this work, we propose to investigate the use of Spike Timing Dependent Plasticity (STDP) [2] within SNNs for effectively analyzing underwater acoustic data. Our objective is to develop a representation learning method that requires minimal training data and enables energy-efficient data processing on UUVs in challenging underwater environments.

Underwater Acoustic Data Representation Learning with Spike Timing Dependent Plasticity


COSMOS is a state-of-the-art government cloud infrastructure. See firsthand how its capabilities can support DoD customers, boost efficiency, propel AI/ML adoption forward, and drive mission success.

COSMOS: Moving Mission Light Years Ahead

Intelligent autonomous systems (IAS) are crucial for enabling the Department of Defense (DoD) to operate effectively in complex and contested environments while reducing risks to personnel and mission success. Maritime domain awareness (MDA) can be significantly enhanced by deploying numerous, distributed, and attritable autonomous platforms capable of adapting to dynamic, constrained, or denied environments.

This research explores autonomous, multi-agent solutions for maintaining resilient communications in these types of scenarios. Our approach focuses on dynamically positioning multiple heterogeneous unmanned platforms to optimize mesh network connectivity in order to maintain connectivity with the fleet. We employ multi-agent reinforcement learning (MARL) algorithms to jointly maximize search effectiveness and network link strength.

JAZZ: Autonomous Multi-Agent Systems for Maritime Network Coordination and Resiliency

This talk is going to discuss recent advances and next steps for assured spectrum perception using AI/ML techniques in tactical environments.

Resilient Dynamic AI/ML for Assured Spectrum Perception in Tactical Environments: Recent Advances, Existing Challenges and Next Steps

Understanding the agent’s learning process, particularly the factors that contributeto its success or failure post-training, is crucial for comprehending the rationalebehind the agent’s decision-making process. Prior methods clarify the learningprocess by creating a structural causal model (SCM) or visually representing thedistribution of value functions. Nevertheless, these approaches have constraintsas they exclusively function in 2D-environments or with uncomplex transitiondynamics. Understanding the agent’s learning process in complex environmentsor tasks is more challenging. In this paper, we propose REVEAL-IT, a novelframework for explaining the learning process of an agent in complex environments.Initially, we visualize the policy structure and the agent’s learning process forvarious training tasks. By visualizing these findings, we can understand how mucha particular training task or stage affects the agent’s performance in the test. Then,a GNN-based explainer learns to highlight the most important section of the policy,providing a more clear and robust explanation of the agent’s learning process.The experiments demonstrate that explanations derived from this framework caneffectively help optimize the training tasks, resulting in improved learning efficiencyand final performance

REVEAL-IT: REinforcement learning with Visibility of Evolving Agent poLicy for InTerpretability

Station-Keeping short-duration high-altitude balloons (HABs) in a region of interest is a challenging path-planning problem due to partially observable, complex, and dynamic wind flows. Deep reinforcement learning is a popular strategy for solving the station-keeping problem. A custom simulation environment was developed to train and evaluate Deep Q-Learning (DQN) for short-duration HAB agents in the simulation. To train the agents on realistic winds, synthetic wind forecasts were generated from aggregated historical radiosonde data to apply horizontal kinematics to simulated agents. The synthetic forecasts were closely correlated with ECWMF ERA5 Reanalysis forecasts, providing a realistic simulated wind field and seasonal and altitudinal variances between the wind models. DQN HAB agents were then trained and evaluated across different seasonal months. To highlight differences and trends in months with vastly different wind fields, a Forecast Score algorithm was introduced to independently classify forecasts based on wind diversity, and trends between station-keeping success and the Forecast Score were evaluated across all seasons.

Seasonal Station Keeping of Short-Duration HABs using Deep Reinforcement Learning (DRL)

Knowledge of underwater acoustics in diverse oceanic environments is key to accurate analysis of sonar systems used for Air Anti-Submarine Warfare (ASW) simulations. By identifying variations in underwater sound propagation, analysts score the acoustic abilities of sonar tools according to their performance in distinct regions. Since these variations are sensitive to environmental changes, leveraging known underwater properties to group oceanic regions could be used to improve acoustic analysis and inform decision makers.

Ocean Provincing Tool

Digital twins provide real-time, high-fidelity predictions that are essential for fatigue
analysis, assessing the integrity of components under extreme loading, and designin
resilient materials for aerospace, automotive, and civil infrastructure. Accurate modeling of material response to deformation is critical to ensure safety and performance across industries. For this purpose, machine learning (ML) has emerged as a promising alternative to physics-based simulations, which often rely on constitutive relations that do not fully capture all processes present in real-world scenarios. However, ML models lacking  physically principled design struggle to generalize effectively, particularly in scenarios involving complex strain histories and multiaxial loading conditions.

This work demonstrates the advantage of enforcing material objectivity, a fundamental physical constraint that governs how the stress tensor transforms under rotations and reflections. We compare a standard convolutional neural network to an equivariant convolutional neural network, which exactly respects O(3) symmetry at the architectural level. Both models are trained on the strain to stress relationship of J2 plasticity; however, only the equivariant model is guaranteed to learn mappings consistent with the principle of material objectivity. As a result, our equivariant model achieves superior prediction accuracy across complex deformation scenarios.

Learning Metal Plasticity with an Equivariant Neural Network

The convergence of technologies—from foundation AI models to diverse sensors and actuators to ubiquitous connectivity—is transforming the nature of interactions in the physical and digital world. People have accelerated their collaborative connections and productivity through digital and immersive technologies, no longer limited by geography or language or access. Humans have also leveraged and interacted with AI in many different forms, with the advent of hyperscale AI models (i.e., large language models) forever changing (and at an ever-astonishing pace) the nature of human-AI teams, realized in this era of the AI “copilot.” Similarly, robotics and automation technologies now afford greater opportunities to work with and/or near humans, allowing for increasingly collaborative physical robots to dramatically impact real-world activities. It is the compounding effect of enabling all three capabilities, each complementary to one another in valuable ways, and we envision the triad formed by human-robot-AI teams as revolutionizing the future of society, the economy, and of technology.

Premium content

Next from NAML 2025

Automated Aircraft Sidenumber Recognition using Computer Vision for Naval Aircraft Logistics

Similar lecture

Real or Robotic? Assessing Whether LLMs Accurately Simulate Qualities of Human Responses in Human-LLM Dialogue

Stay up to date with the latest Underline news!

PRESENTATIONS

CONFERENCES

COMPANY

RESOURCES