Digital Commons Network^™

Framework For Process-Aware Cyber-Physical Security Monitoring, Detection, Triage, And Response, Moath Sakaji

Defensive Publications Series

This disclosure describes techniques of process-aware cyber-physical security in an operational technology (OT) environment. Time-series process data is ingested from a secure replica of a process historian and IT security data. A cyber-physical context engine (CPCE) includes a library of machine learning models each trained for a specific operational state. The CPCE analyzes the process data to detect physical anomalies. Physical anomalies are correlated with IT security data to identify cyber-physical attack patterns. An incident response module enforces a safety-gated, collaborative workflow between security analysts and plant operators, prioritizing physical process stability. A generative AI-assisted workflow enables human analysts to …

Ai Context Automaton – Extended Architecture For Compliance-Aware Local Memory And Environmental Optimisation, David R P Chiesa

Defensive Publications Series

AI Context Automaton introduces a dual-AI architecture that transforms stateless AI interactions into a stateful, compliance-aware, and environmentally optimised workflow. The system combines a local AI for memory management and compliance enforcement with a cloud-based AI for reasoning, leveraging a markdown-based knowledge base for persistent storage. By offloading memory operations and compliance checks to local devices, the architecture minimises repeated cloud compute cycles, reducing energy consumption, carbon footprint, and operational costs while ensuring adherence to organisational governance standards.

Ai-Based Digital Concierge For Complex Financial Queries, Ankit Aggarwal, Jose Estevez, Chin Powit Rungsangthiwakorn, Samreen Dhillon

Defensive Publications Series

Maximizing savings, e.g., by choosing an appropriate form of payment, voucher/offer, merchant, etc. is difficult due to the need to analyze a large amount of information to determine the best combination when making a purchase. This disclosure describes a digital payment concierge that can answer complex financial queries. The concierge is implemented using an agent-based framework. To understand user intent, an orchestration agent is implemented using a suitable artificial intelligence model and a custom prompt. The orchestration agent routes tasks to specialized subagents, such as a savings maximizer which finds the best price by stacking offers and considering the user's …

System And Method For Automatically Selecting Context-Aware Llms Using A Reward-Based Multi-Objective Learning Mechanism, Sumit Kumar, Shivam Mohan, Ravi Shanker Kumar Sinha, Dheeraj Dhamija, Soumendra Kumar Mishra

Defensive Publications Series

The present disclosure provides a method and a system for automatically selecting context-aware large language models (LLMs) using a reward-based multi-objective learning mechanism. The method includes receiving an input task and extracting a corresponding context vector comprising one or more contextual attributes. The method includes computing, for each of a plurality of LLMs, a confidence-adjusted expected performance score using the context vector, the score being generated using a contextual learning algorithm. The method includes selecting at least one of the plurality of LLMs having the highest confidence-adjusted expected reward for the input task. The method includes executing the selected LLMs …

Dynamic Generation Of Contextual User Interface Controls From Content Analysis, Sönke Rohde, Alan Ross, Carsten Hinz

Defensive Publications Series

User interfaces for content generation and manipulation can be limited by static controls or ambiguous inputs that may not provide precise, context-specific tools. The disclosed technology relates to a system that can programmatically generate adaptive user interface (UI) controls based on content analysis. This system can analyze a corpus, such as an image or document, to perform semantic inference and derive manipulable properties. These properties can be mapped to declarative UI definitions for controls, such as sliders or dropdown menus, which can be rendered by a host application. This process can provide a user with context-aware controls tailored to specific …

The Ψ-Theory: Unifying Cosmology And Quantum Mechanics Through A Dynamic Temporal Metric, Valerii Y. Vorobiev

Defensive Publications Series

We propose the Ψ-Theory, a novel ontological framework that addresses fundamental crises in modern cosmology by redefining the nature of time and gravity. The theory posits a primary scalar (or, fundamentally, tensorial) field Ψ, which is not a traditional force field, but a metric of the local density of causal interactions (quantum "Acts") in the network of reality. From this field, the observable geometry of spacetime and the phenomenon of gravity emerge as secondary, effective descriptions. The core principle is that time is not a background entity, but a dynamic, relational metric of causal connectivity. The Ψ-field dictates the local …

Novel Crystalline Forms Of 4-[3­Chloro-4-(N’-Cyclopropylureido)Phen Oxy]-7-Methoxyquinoline-6-Carboxamide Ethanesulfonate And Process For The Preparation Thereof, Srinivasan Thirumalai Rajan, Sagyam Rajeshwar Reddy, Kammari Balraju, Kamutam Sathish, Msn Laboratories Private Limited, R&D Center, Hyderabad, India.

Defensive Publications Series

Abstract:

The present disclosure relates to a novel crystalline forms of 4-[3­chloro-4-(N’-cyclopropylureido)phenoxy]-7-methoxyquinoline-6-carboxamide ethanesulfonate which is represented by the following structural formula-1b, which is referred to as Lenvatinib ethanesulfonate.

Formula-1b

Reconstructing Missed Photographic Moments Within A Spatiotemporal Event Cluster, Ilia Labzovsky, Danny Karmon

Defensive Publications Series

Group photos captured at events such as birthdays, weddings, holidays, etc. may sometimes leave out important individuals, rendering the memory of the event incomplete. This disclosure describes techniques that utilize generative artificial intelligence to reconstruct missed moments within a set of photos taken during a particular event by leveraging the collective context of the photos within the photo set. A multi-state artificial intelligence (AI) pipeline is implemented that includes missed moment identification using event analysis; asset and scene reconstruction; context-aware reconstruction; and generative rendering. The techniques shift the paradigm of digital photo albums from organizing photos to helping users obtain …

Llm-Based Automatic Generation Of Contextually Appropriate User Interface, Ilia Labzovsky, Danny Karmon

Defensive Publications Series

Currently, the user interface (UI) for digital photo management applications is a static, reverse chronological grid of thumbnails. From a user perspective, this UI has several drawbacks such as high interaction cost for retrieval, lack of contextual adaptation, and higher task friction. This disclosure describes a generative situational UI engine for media management applications. The engine is powered by an integrated large language model that is capable of automatic user interface generation. This engine replaces the static, one-size-fits-all grid with a dynamic interface that, with the user’s permission, is generated based on the user’s context and the content of their …

Coordinated, Heterogeneous Model Composition Under A Shared Governance And Audit Layer, Joseph Jm Walker

Defensive Publications Series

This paper introduces the Modular Artificial Specialized Intelligence (MASI) framework, a distributed cognitive architecture designed to address the safety, transparency, and governability limitations of monolithic foundation models. MASI decouples reasoning into four canonical, independently operational roles—Wisdom, Foresight, Empathy, and Precision—coordinated through a standardized communication layer known as the MASI Bus. The framework enforces a Consult-Before-Execute (CBE) protocol, requiring multi-perspective verification, drift detection, and ethical arbitration prior to any system output or tool execution. This deliberative process is quantified via the MASI Trust Score ($T_{masi}$), a dynamic metric integrating fairness, logic, and uncertainty into a verifiable audit artifact. Additionally, the …

Model-Driven Continuous Validation Using Adversarial Testing In A Digital Twin, Ramakrishnan Neelakandan, Teginder Singh, Bakul Patel

Defensive Publications Series

Conventional software validation methods can be static and periodic, which may present challenges for continuously updated platforms (e.g., cloud platforms) in regulated industries and can impede the validation of non-deterministic systems like artificial intelligence models. A system for continuous validation may use a dual-pathway architecture guided by a machine-readable compliance model. One pathway can perform adversarial testing within an isolated digital twin of a production application to discover potential compliance weaknesses. Concurrently, a second pathway can provide real-time observational monitoring of the live production system for policy deviations and anomalies. Findings from both pathways may be consolidated into a persistent, …

Multimodal Ai For Descriptive And Causal Analysis Of Infrastructure Defects From Imagery, Theodorus Tjhang

Defensive Publications Series

Some automated systems for infrastructure inspection may detect defects from imagery but could lack the ability to provide descriptive context, causal reasoning, or interactive querying. A system is described that can utilize a multimodal artificial intelligence model to analyze visual data, such as pre-existing geolocalized imagery, in conjunction with natural language text prompts. The system can generate detailed analytical outputs including, for example, segmentation masks and bounding boxes for defect localization, as well as descriptive labels detailing the nature of a defect, its potential causes, and the reasoning for its detection. This approach can facilitate a more comprehensive and interactive …

Self-Improving Autonomous Data Exploration System With Meta-Learning And Temporal Knowledge Graph, Sai Charan Tej Kommuri, Madhup Sukoon

Defensive Publications Series

Data exploration methods that rely on manual intervention may be limited in their ability to autonomously discover insights from large, unanalyzed enterprise datasets. To address this, systems and methods are described that can employ a self-improving, multi-agent architecture. This system can feature an exploration layer for generating hypotheses, an evaluation layer for validation, a temporal knowledge graph (TKG) to serve as a persistent memory of insights, and a meta-learning layer. The meta-learning layer can observe analytical outcomes stored in the TKG to refine future exploration policies. This closed-loop process can enable the system to autonomously generate, validate, and learn from …

[Conversational Operating System] Dynamic System Prompt Architecture For Context State Governance In Large Language Models, Joseph Jm Walker

Defensive Publications Series

Core Insight

The system prompt should not be treated as static instructions.
It should be treated as a dynamic operating layer that governs reasoning, tone, epistemic discipline, bias correction, and long-term continuity.

What Makes This Distinct

Most prompt engineering focuses on what the model says.
This work focuses on how the model reasons, under what constraints, and how that behavior persists over time.

This disclosure presents a dynamic system-prompt architecture that transforms the system prompt of a large language model from a static instruction block into an adaptive governance layer. The architecture introduces a context-state machine, an operator-level control protocol, …

Process For The Preparation Of Palbociclib, Msn Laboratories Private Limited, R&D Center; Srinivasan Thirumalai Rajan; . Irukuvajjula Venkata Subrahmanya Vithal, Kesari Sai Manoj Krishna And Bhatraju Daveedu

Defensive Publications Series

The present disclosure provides a process for the preparation of controlled particle size and specific surface area of crystalline form-A of Palbociclib and its process for the preparation. Palbociclib is represented by the following structural formula-1.

Formula-1.

Process For The Preparation Of Gadolinium-Based Contrast Agents And Its Intermediates, Msn Laboratories Private Limited, R&D Center; Srinivasan Thirumalai Rajan; Sagyam Rajeshwar Reddy; Velishala Praveen And Vadlakonda Srikanth.

Defensive Publications Series

The present application relates to a process for the preparation of 2,2',2''-(1,4,7,10-Tetraazacyclododecane-1,4,7-triyl)triacetic acid represented by the following structural formula-3.

Formula-3.

Present disclosure also describes a novel crystalline form of Calteridol calcium and its process for the preparation.

Constrained Response Generation For Conversational Generative Ai Agents, Łukasz Bieniasz-Krzywiec, Kyle Boone, Ryan Carl Yao Yu

Defensive Publications Series

The use of generative artificial intelligence (AI) agents in various applications can be affected by the underlying models exhibiting hallucinations and/or vulnerability to prompt injection attacks. In customer-facing applications, such as taking orders for quick service restaurants (QSR) or drive-through restaurants, model hallucinations can create a risk of brand damage and/or bad publicity. This disclosure describes techniques that constrain the output of generative AI agents using deterministic control to reduce off-topic or inappropriate responses in structured conversational contexts such as order-taking in a quick service restaurant. A deterministic control module includes a state machine whose states are the current context. …

Ai-Assisted Meeting Lifecycle Management For Enhanced Productivity, Anton Volkov, Lindsay Moore, Kimberly Ha, Dom Ekeledo, Juan Angustia, Felix Mejia, Sara Olson, Joy Barlow, Awaneesh Verma, Avery Fisher, Adam Karnas, Ravi Mruthyunjaya, Rabia Murtza, Kristin Moore, Kelly Dern, Yury Pinsky, Ryan Fedyk, Arvid Callas

Defensive Publications Series

Meeting productivity is often hindered by attendees being unprepared, a lack of clear objectives, and difficulty in tracking discussions and outcomes. This leads to inefficient use of time and a loss of valuable information, as post-meeting actions and decisions are frequently unclear or forgotten. The problem extends beyond video conferences to phone and in-person meetings, which often lack any formal support structure. This disclosure describes a system that uses artificial intelligence to manage the entire meeting lifecycle. The system provides pre-meeting briefings with contextual information, offers real-time in-meeting assistance for note-taking and agenda management, and generates post-meeting summaries with actionable …

Cross-Probe Virtual Element Synthesis For Ultrasound Imaging, Diane Manthey, Anonymous

Defensive Publications Series

This invention describes a system and method for synthesizing virtual ultrasound elements by combining physical elements from two separate probes

Adaptive Placement Of Augmented Reality User Interface Elements Using Reinforcement Learning, Dongeek Shin

Defensive Publications Series

Conventional augmented reality (AR) systems often struggle to accurately and contextually place user interface (UI) elements, leading to a non-immersive or intrusive user experience. This disclosure presents a method for personalized UI placement in AR through reinforcement learning. The technique utilizes an agent trained with deep reinforcement learning to determine the optimal position and orientation of virtual UI elements. Input features, including environmental depth maps, eye-tracking data, and application context, are processed to inform placement decisions. The agent learns from both implicit user interactions, like gaze aversion or physical interference, and explicit user corrections. This process allows the UI placement …

Cross-Device Eye-Tracking Calibration Via Reference Model Remapping, Mar Gonzalez Franco, Eric Jordan Gonzalez, Lucy Abramyan, Karan Ahuja, Prasanthi Gurumurthy, Khushman Jayantilal Patel

Defensive Publications Series

Eye-tracking calibration models are often specific to a particular device’s sensor geometry, requiring extensive data collection and retraining when camera or LED positions are altered in new hardware. This process is resource-intensive and hinders the portability of models across different devices. A method is disclosed for adapting a comprehensive eye-tracking model from a reference device to a new device with a different sensor layout. By collecting a minimal set of new calibration points on the target device, a transformation matrix is generated. This matrix remaps the existing reference model to the new geometry. The primary purpose of this technique is …

Visual Cue For Item Value Discrepancy In An Augmented Reality Environment, Nick Dugal

Defensive Publications Series

Wearable devices, such as augmented reality (AR) devices and smart glasses, lack an intuitive, non-disruptive mechanism for informing a user about suboptimal pricing for a viewed item. This disclosure describes a method for displaying a dynamic visual cue, such as a stylized line, proximate to an item’s displayed price. The properties of this cue, including color, thickness, or animation, can be modified to convey the magnitude of the value discrepancy. User interaction with the cue can reveal detailed comparison pricing sourced from various databases. This price-matching information can also be transferred to a secondary device. The primary purpose is to …

Automated Spatial Localization Of Iot Devices Via Visual Detection And Network-Initiated Feedback, Michael Christopher Yu

Defensive Publications Series

Manually registering the spatial positions of numerous Internet of Things (IoT) devices within an Extended Reality (XR) environment is a cumbersome process for users. This lack of automated spatial mapping limits the ability to perform context-aware interactions with these devices. The disclosed method addresses this by using an XR device’s vision and depth sensors to first detect a potential IoT device and estimate its 3D position. The XR device then sends a network-based request to candidate devices, instructing them to provide a confirmation signal, such as illuminating an LED. When the XR device’s camera detects this signal, the device's identity …

Design-Time Preview System For Blended User Interfaces On Transparent Displays, Adhithya Ramakumar, Louis Benedict Pullen-Freilich

Defensive Publications Series

Designing user interfaces for transparent additive displays presents a challenge in which graphical blend modes alter specified UI colors based on the real-world background, creating a mismatch between the design intent and the final user experience. This limitation forces a slow, iterative workflow of deploying to a physical device to verify visual appearance. A method is disclosed for generating an accurate design-time preview of such UIs directly within a development environment. The technique renders the complete UI to an offscreen buffer and then composites the UI against a representative background using a graphics layer with a specified blend mode. This …

Predicting User Interaction Intent For Adaptive Ui Highlighting In Extended Reality Environments, Tushar Chugh, Aditya Shrikant Mone, Karthik Kumara

Defensive Publications Series

In extended reality (XR) environments, persistent highlighting of user interface (UI) elements based on gaze can distract users during non-interactive tasks, such as reading. Conversely, a complete lack of highlighting can reduce interaction precision and increase the risk of accidentally activating critical controls. The disclosed technology is a method for adaptive gaze highlighting that dynamically controls the visibility of these visual cues. The method utilizes a model to analyze multimodal inputs, such as gaze patterns, gestural cues, and application context, to predict a user’s intent to interact. A necessity score is calculated based on the UI element’s characteristics and the …

Somatic Control Framework For Tokamak Plasmas, Samuel Church

Defensive Publications Series

This technical design document proposes the Somatic Control Framework, a novel hierarchical control architecture designed to stabilize tokamak plasmas through resonant multi-frequency modulation. Unlike conventional black-box controllers that are often device-specific, the Somatic Framework decomposes plasma dynamics into physically interpretability modes: the Global Heartbeat (M_B) targeting tearing modes, the Edge Theta (M_\theta) managing pedestal stability, and the Fast Gamma (M_\gamma) addressing turbulence.

This entry details the methodology for cross-device transfer, specifically enabling the portability of control policies from DIII-D to KSTAR. We define a scaling law based on dimensionless parameters (including …

Pharmaceutical Composition Of Resmetirom Tablets

Defensive Publications Series

ABSTRACT

A stabilized pharmaceutical composition comprising Resmetirom or a pharmaceutically acceptable salt thereof and at least one pharmaceutically acceptable excipient, wherein said composition simultaneously achieves desired dissolution profile, stability and bioavailability.

Distributed 48 V Energy Nodes (D48en), Hieu Nguyen

Defensive Publications Series

This document introduces the Distributed 48 V Energy Node (D48EN) architecture, a modular approach to home and community-scale energy storage and power distribution. Unlike conventional DC microgrids that rely on a continuous shared bus, this architecture places independent 48 V battery nodes throughout a structure, each with its own point-of-load photovoltaic (PV) input and short-run DC distribution to nearby loads. Nodes are normally isolated electrically and do not form a high-power backbone. Instead, low-current redistribution links allow controlled, schedulable energy sharing between nodes. This configuration reduces wiring cost, improves safety, simplifies retrofits, and enables modular growth for both developed residential …

Single Pointer Operation To Rearrange External Display Positions, Shaun Tollerton, Nergi Rahardi

Defensive Publications Series

Many computing devices support multiple displays, with features that enable rearranging the positions of the displays to match their respective physical positions. Such rearrangement typically uses hold-and-drag operation performed via a mouse, touchpad, or equivalent. This mechanism can be problematic for certain users and/or unavailable when the available input devices do not support hold-and-drag. This disclosure describes techniques that enable a user to rearrange multiple displays without needing to perform hold-and-drag or other operations that are infeasible. Per the techniques, the user first provides a selection input in a user interface to select a particular display and subsequently provides additional …

End-To-End Differentiable Gnss Positioning With Neural Network-Derived Weights, Yu Liu, Duong Hoang, Ian Atkinson, Mohammed Tawashi, Saket Patkar, Anant Sahai

Defensive Publications Series

There is a fundamental limitation in accuracy and robustness of traditional Global Navigation Satellite System (GNSS) positioning techniques that rely on the Weighted Least Squares (WLS) algorithm. Traditional techniques use static heuristic functions for assigning weights to pseudorange measurements in the WLS algorithm. This disclosure addresses these limitations by replacing the static heuristic functions that are used for assigning weights to pseudorange measurements in the WLS algorithm with a trained Neural Network (NN). This NN acts as a dynamic Satellite Weight Prediction Model. Per the techniques described herein, the entire positioning pipeline—from satellite features to final positional error—is treated as …