Digital Commons Network^™

Heart-Rate-Recovery-Gated Rest-Interval Timing With Audio Status Updates Through A Consumer Earbud, Gerard Ziemski

Defensive Publications Series

This disclosure describes a method of dynamically gating the rest interval between exercise rounds based on real-time heart-rate recovery rather than a fixed-duration timer, with audio feedback announcing the remaining recovery distance through a consumer earbud.

At the end of each round the system records the round-end heart rate and computes a personalized recovery target as a percentage of an age-derived maximum heart rate. A minimum rest duration prevents excessively short rests when heart rate drops rapidly; a maximum rest duration prevents excessively long rests when heart rate does not recover. The next round begins when either (1) the current …

Consumer Earbud As A Closed-Loop Biometric Sensor And Ambient Feedback Device For Exercise Support, Gerard Ziemski

Defensive Publications Series

This disclosure describes the use of a single consumer earbud — an in-ear device with an integrated optical (photoplethysmographic) heart-rate sensor — simultaneously as both the heart-rate sensor and the sole audio feedback channel for real-time, hands-free, screen-free exercise support, without any separate wearable such as a smartwatch or chest strap.

Heart-rate and related samples (active energy, respiratory rate, step count, distance, flights climbed) are obtained through the host operating system's on-device health and workout framework, keeping all biometric computation on the device. A heart-rate zone configuration is derived from age (maximum heart rate estimated as 208 minus 0.7 times …

Crediting Recorded Prior Activity Toward An Exercise Warm-Up, With Heart-Rate-Gated Warm-Up Termination And Audio Status Through A Consumer Earbud, Gerard Ziemski

Defensive Publications Series

This disclosure describes a method for shortening or waiving the warm-up period before an exercise session by crediting the user's recorded prior activity, in combination with heart-rate-gated warm-up termination and audio status delivered through a consumer earbud.

A fitness application queries a platform health-data store (for example Apple HealthKit or Google Fit) and/or its own session log for activity bouts ending within a recency window before the present session — a previously recorded workout, or elevated active-energy or heart-rate samples. When a qualifying bout is found, the minimum warm-up duration is reduced or waived, with the amount of credit scaling …

Dynamic Video Resolution For Background Playback, Daniel Vik, Jugal Manjeshwar

Defensive Publications Series

This document describes techniques for a context-aware media scaling system that dynamically adjusts the resolution for video playback on devices (e.g., resource-constrained devices such as smart televisions) based on the current state and navigational context of a user interface (UI). The system determines the current state by actively tracking an application’s view context or page identifier (ID) to detect navigational transitions between a primary viewing UI (e.g., a “Watch” page) and a secondary navigational UI (e.g., for browsing content). The system maps the current state to a specific resolution policy. For example, for a transition from the primary viewing UI …

Calibration Of A Quantum Random Number Generator System Using Machine Learning, Ashish Kundu, Ramana Kompella, Peng Zhao, Eneet Kaur

Defensive Publications Series

Disclosed are techniques for calibrating a quantum random number generator (QRNG) system using machine learning to detect noise, drift, and entropy degradation across devices and the QRNG pipeline. One or more models evaluate device data, entropy quality, and system conditions to determine whether recalibration is needed, which devices should be recalibrated, and when recalibration should occur. The techniques may also support predictive recalibration and calibration planning to improve entropy quality and overall QRNG reliability.

Autonomous Multi-Agent System For Declarative Data Center Lifecycle Management, Ananya Simlai, Niccolò Cascarano

Defensive Publications Series

Conventional data center management can involve manual workflows or static declarative systems that may not effectively adapt to real-time conflicts or complex events, potentially resulting in unpredictable deployment timelines and increased operational risk. Systems and methods are described that can utilize a multi-agent ecosystem for managing data center resources. In such a system, specialized, autonomous software agents can communicate and negotiate with one another through a medium such as an asynchronous fabric to interpret high-level declarative goals. The agents may autonomously orchestrate complex actions, resolve resource conflicts, and adapt to changing conditions. This approach can improve operational resilience and agility …

Adaptive Video Sampling Using Multi-Factor Analysis For Multimodal Ai, Larry Lai, Ruixi Fan, Lan Wang, Youbao Tang

Defensive Publications Series

Processing continuous video streams can be computationally expensive and may lead to information loss with fixed-rate sampling techniques. A content-aware filtering system for adaptive video sampling may utilize a set of parallel analysis components to examine a video stream in real-time for factors such as scene changes, motion, semantic content, and latent feature similarity. Based on the aggregated output of these components, the system can select information-rich keyframes for downstream processing. A selected keyframe may be paired with a temporal timestamp. This approach may reduce data volume and computational load compared to fixed-rate sampling, while seeking to preserve temporal details …

Artificial Intelligence-Assisted Audio Enhancement And Synchronization In Smartphones For Hearing Aids, Pranav Kuriakkattil

Defensive Publications Series

This publication describes systems and techniques for providing artificial intelligence (AI) assisted audio enhancement and synchronization for hearing aids using an associated smartphone to improve audio quality available to the hearing aid user.  The system overcomes the hardware limitations of conventional hearing aids and improves the user experience, by utilizing the processing power, sensors, and connectivity of the user’s smartphone. The system creates a distributed microphone array by combining hearing aid microphones with smartphone microphones and offloading signal processing to the smartphone to preserve hearing aid battery life. Utilizing AI models for environment classification, speaker tracking, and binaural synchronization, the …

Real-Time Generative Audio Description Insertion Using Vision Language Analysis, Nikita Agarwal

Defensive Publications Series

Standard audio description tracks are manually scripted and pre-recorded, leaving vast amounts of uncurated media content inaccessible to visually impaired users as the audio description tracks are only available for a small fraction of premium media content. To address this limitation, the disclosed technology details a process for generating real-time audio descriptions for media content. Video frames are analyzed locally by an on-device vision language model during identified silence windows or gaps within the media content. A text summary of the visual action is generated and subsequently converted to speech. The playback speed of the synthesized narration is dynamically adjusted …

Adaptive Compliance-Aware Secure Gateway Selection And Disaster-Aware Autonomous Tunnel Rehoming For Secure Service Edge (Sse) Environments Using No-Mutation Shadow Probing And Policy-Equivalence Gating, Faizan Amjad Mohammed, Ashwin Kumar Ravichandran, Pradeepan K Kannawadi

Defensive Publications Series

Proposed herein is a framework to enable adaptive, compliance-aware secure service edge (SSE) secure gateway selection and autonomous tunnel rehoming for enterprise software-defined wide area network (SD-WAN) edges in both steady-state and disaster conditions. The proposed framework uniquely combines no-mutation controller-side shadow probing, hard policy-equivalence gating, and transactional canary/staged rollout with deterministic rollback before production cutover. The result is safer failovers and better application performance while preserving compliance boundaries and reducing outage blast radius.

Dynamic Health-Aware Agent Card Extensions To The Agent-To-Agent (A2a) Protocol For Fitness-Based Orchestration In Multi-Agent Systems, Robert Barton, Yathi Udupi, Maithri V M, Jerome Henry

Defensive Publications Series

A service is proposed herein that updates Agent-to-Agent (A2A) cards in real-time regarding the health and performance of a particular agent, including information represented in the particular agent's card, so that other agents have better information before selecting the particular agent for a given task/operation.

Improved Process For The Preparation Of Retatrutide, Anonymous

Defensive Publications Series

The present invention relates to an improved process for the preparation of Retatrutide or a pharmaceutically acceptable salt thereof. The present invention also relates to novel fragments as intermediates and use thereof in the preparation of Retatrutide or a pharmaceutically acceptable salt thereof.

Solid State Form Of Resmetirom And Its Processes For Preparation, Msn Laboratories Private Limited, R&D Center; Hyderabad, India; Srinivasan Thirumalai Rajan; Revu Satyanarayana, Nomula Sathaiah And Rayavarapu Srinuvasarao.

Defensive Publications Series

The present disclosure relates to a novel crystalline form of 2-[3,5-dichloro-4-(5-isopropyl-6-oxo-1,6-dihydropyridazin-3-yloxy)phenyl]-3,5-dioxo-2,3,4,5-tetrahydro[1,2,4]triazine-6-carbonitrile represented by the following structural formula-1, which is referred to as Resmetirom.

Formula-1.

Audio Signal Correction Via Transformer-Based Fusion Of Facial Landmarks And Latency Metadata, Ruofei Du, Mahdi Tayarani Najaran

Defensive Publications Series

In remote communication, audio signals and visual signals are often unsynchronized. A mismatch between perceived visuals and corresponding audio leads to speech recognition errors and reduced audio quality. This disclosure provides an architecture for visual-audio speech correction. Features are extracted by a mouth landmark encoder, an audio encoder, and a visual encoder. These features are mapped to a target audio signal through an on-device visual-audio transformer. Simultaneous processing of signals from both visual and audio modalities is performed. Device latency and conferencing platform latency are integrated into transformer encoders during a preprocessing stage. Enhanced audio signals are then decoded to …

Predictive Intent Proposal And Gestural Confirmation In Head-Worn Displays Via Multimodal Large Language Model Analysis, Ilia Labzovsky, Tamar Yakar

Defensive Publications Series

Traditional interaction models for wearable displays often present challenges such as physical fatigue, social friction, and unintended activations. Fully autonomous execution based on inferred intent may also lead to high-stakes errors. This disclosure describes a method for predictive intent proposal and non-verbal confirmation using multimodal data analysis. Video streams, eye-vector data, and inertial measurement unit data are ingested. Raw eye movements are converted into semantic tokens. An on-device multimodal large language model processes these tokens and the visual scene to infer a likely user intent. A non-intrusive augmented reality overlay is generated to propose a specific action. The action is …

Augmented Reality Glasses For Monitoring External Environments Via Internal Reflections, Michael Christopher Yu

Defensive Publications Series

Existing vehicle safety systems often involve expensive, factory-integrated camera arrays and sensors that are difficult to upgrade and susceptible to occlusion from environmental debris, such as dirt or rain. Many vehicles possess basic backup cameras, lacking comprehensive advanced driver assistance capabilities.

This publication describes a method for providing driver assistance through a wearable extended reality device. The device utilizes inside-out spatial registration to identify the location of statically-positioned vehicle mirrors within the interior.

Images of the mirrors are captured and subsequently processed using an image cropping mechanism. Computer vision is then applied to these isolated segments to detect the presence …

Interactive Picture-In-Picture Control Surface For Managing Background Agentic Tasks, Yijo Liu, Clifford Tse Yan Chan, Tiffany Deonne Morris, Shadia Walsh, Golden Gopal Krishna, Lily Duraiswami, Yeun Su Shin, Carl Magnus Borg

Defensive Publications Series

Background processes are frequently suspended by mobile operating systems, which obstructs multitasking during long-running automated browser workflows. Visual feedback regarding the execution state of these workflows is typically absent. This disclosure explains how a live visual preview of an active Document Object Model is converted into a data stream compatible with a picture-in-picture container. A floating window is generated when a context switch occurs away from a mobile browser. A text summary is superimposed on the container to display specific workflow states. Interactive prompts are included directly within the overlay to allow state modifications. User intervention is enabled without a …

Interactive Graphical Interface For Multi-Tab Content Synthesis And Parametric Control, Yijo Liu, Carl Magnus Borg, Tiffany Deonne Morris, Shadia Walsh, Golden Gopal Krishna, Melissa Kuo Bui, Yeun Su Shin, Clifford Tse Yan Chan, Lily Duraiswami

Defensive Publications Series

Current methods for comparing content across multiple browser tabs are limited because tabs are treated as separate rendering instances. Comparisons of data across different pages often involve manual navigation or reliance on static text tables that lack interactivity. A technique is disclosed where a comparison interface is generated by extracting attributes from the underlying data structures of selected browser tabs. This interface synthesizes extracted data into a graphical overlay including parametric controls such as sliders and dials. Interactions with these controls trigger modifications to the browser state. These modifications include the physical reordering of tabs within a tab strip, visual …

Gating Spatial Coordinates Utilizing Inertial Detection, Tushar Chugh, Aditya Shrikant Mone, Karthik Kumara

Defensive Publications Series

Extended reality (XR) input systems relying solely on optical hand-tracking face critical limitations, including contact ambiguity between hovering and touching, processing pipeline latency, and field-of-view occlusion. This disclosure describes a multimodal sensor fusion framework that addresses these challenges by integrating spatial coordinate mapping from head-worn devices with high-frequency inertial event detection from wrist-worn devices. When a wrist-worn sensor detects the sharp acceleration of a physical impact at a specific timestamp, the system triggers retroactive event gating. The system accesses historical visual buffers to retrieve spatial coordinates corresponding to the impact time, then applies an update that dynamically forces the depth …

Generative Pipeline For 2d Image To Volumetric 3d Media Conversion In Extended Reality Environments, Dev Shimulbhai Daftari

Defensive Publications Series

The adoption of extended reality (XR) headsets is limited by a lack of native volumetric content, resulting in flat presentations of standard two-dimensional media libraries. Traditional depth-mapping approaches introduce severe visual stretching artifacts during six-degrees-of-freedom (i.e., translation and orientation) movement due to missing data behind foreground objects. The disclosed technology introduces a generative artificial intelligence (AI) pipeline that synthesizes volumetric media from standard monocular two-dimensional inputs (e.g., an image or a video frame). The pipeline performs depth estimation from the two-dimensional input to construct a three-dimensional point cloud, computes left and right eye stereoscopic perspectives via head-pose tracking, and uses …

Ai-Driven Application Policy Infrastructure Controller Managed Object Configuration And Validation System, Ramesh Babu Radhakrishnan, Bharathan Thangam Madhavenpillai, Amrit Tickoo, Praveen Kumar Tarikere Nagaraj

Defensive Publications Series

A system is proposed herein that provides an artificial intelligence (AI)-driven autonomous testing framework that programmatically transforms network-controller object schemas into executable validation pipelines. The framework enables systematic and scalable validation of managed objects (MOs) in an intent-based networking system, such as an application policy infrastructure controller (APIC). Large language models (LLMs) interpret structured MO metadata and synthesize hierarchically valid configuration payloads, attribute-complete valid, invalid, and boundary test cases, deterministic verification logic, and cleanup sequences.

Ai Agent Evaluation With Skill-Driven Test Cases And Lightweight Network Simulation, Ryan Holland, Elango Ganesan, Samer Salman, Yao Zhao

Defensive Publications Series

Presented herein is a system for automated generation, deployment, and evaluation of test cases for artificial intelligence (AI)-powered agents performing network operations. The system evaluates long-running agent trajectories rather than only single-turn outputs, validates whether required tool calls were executed, assesses whether a reasonable diagnostic process was followed, and checks whether conclusions were accurate and helpful. Procedural documentation, such as skills, is used as a proxy for coverage so that key use cases can be prioritized and test cases can scale with domain growth. The system also generates and pre-validates synthetic responses to anticipated tool calls, thereby avoiding physical networks …

Ensuring Symmetric Path Forwarding Of Ptp Packets Using A Dedicated Ptp Mac Address Steering Table, Anonymous

Defensive Publications Series

A mechanism is disclosed for maintaining path symmetry of Precision Time Protocol (PTP) message exchanges in a packet switched network. Each network node along a PTP packet path maintains a dedicated PTP Media Access Control (MAC) Address Steering Table that is separate from conventional data plane forwarding and MAC learning tables. PTP packets are identified by User Datagram Protocol (UDP) port 319 for event messages and UDP port 320 for general messages. Upon receiving a forward direction PTP packet from a PTP Grandmaster (GM) toward a Telecom Boundary Clock (T-BC), each node records the source MAC address, source Internet Protocol …

Contextual Hardware Security Using Autonomic Physiological State Signatures, Sudeep Chauhan, Ehsan Nourbakhsh

Defensive Publications Series

This document describes techniques and architectures for securing a computing device against physically coerced access by utilizing dynamic autonomic physiological state signatures. Processing circuitry of a computing device (e.g., a primary computing device, a wearable sentinel device, or an industrial control terminal) continuously determines a dynamic physiological baseline from sensor data (e.g., heart rate, galvanic skin response, inertial data). A physiological baseline and anomaly engine utilizes a machine learning model, which accounts for contextual states such as active exercise tracking to filter false positives, to detect a distress state corresponding to a non-voluntary autonomic deviation relative to the dynamic physiological …

Veti-3.0 (Vortex Ejector Tip Integrated, Michael Victor Caldwell Mr.

Defensive Publications Series

The VETI-3.0 is a passive, bolt-on exhaust scavenging device engineered for naturally aspirated internal combustion engines. Unlike static "vortex cones," this system utilizes Multi-Band Helmholtz Acoustic Tuning to prevent exhaust pulse reversion and an Adjustable Coaxial Ejector to lower effective backpressure via ambient air entrainment. It requires no electronics, no moving parts, and no engine modifications, targeted performance upgrades if this works.

Peak Horsepower Gain: +4% to +7% (Engine-dependent).

• Steady-State Cruise MPG: +5% to +8%.

Operational Range: Optimized from idle to redline.

On-Device Large Language Model For Enhanced User Experience Customization, Yosef Mirsky, Andrew Felch

Defensive Publications Series

A low-intelligence, or small, large language model (LLM), termed a Vibe LLM, operates on resource-constrained hardware to augment user experiences. The Vibe LLM does not handle direct linguistic interaction with a user; instead, it processes textual output generated by a separate, more resource-intensive LLM that typically resides in the cloud. By analyzing this text, the Vibe LLM extracts contextual information, such as proper nouns, and uses it to control device features like light emitting diode (LED) animations or display colors. For example, the Vibe LLM can identify a team name mentioned in a response and trigger an LED animation using …

Ai-Powered Validation Of Screen Reader Output For Automated Accessibility Testing, Shubham Jain

Defensive Publications Series

This technology automates accessibility testing of user interfaces (UIs) by using an artificial intelligence (AI) model to validate screen reader output. The framework uses automation tools to navigate a UI, such as a web page or application, while recording the audio output generated by a screen reader. This initial recording establishes a baseline recording. For subsequent regression testing, the system runs the automation again and compares the new output to the baseline recording to detect any deviations. An AI tool performs the validation by either converting the audio to text for comparison or conducting a direct audio and video comparison. …

Techniques For Preventing Anchoring Bias In Ai Investigative Analysis Through Staged Data Redaction And Hypothesis-Gated Revelation, Avneet Singh Chhabra

Defensive Publications Series

Presented herein is a system for preventing anchoring bias in large language model (LLM) investigative analysis. Bias in LLMs is not typically addressed by instruction-level approaches; rather, it is most effectively mitigated by physically separating the biasing data from an LLM's input during the hypothesis formation phase. The proposed system includes five core components: a Data Redaction Engine, a Staged Analysis Protocol, a Structural Gate, a Corrections Bridge, and a Procedural Attestation Record. The Data Redaction Engine operates to separate source data into a redacted evidence set and an attribution set, validated by a multi-layer validation pass. The Staged Analysis …

Finetuning Multiple Adaptors For Families Of Large Language Models, Ali Payani, Charles Fleming, Ramana Kompella

Defensive Publications Series

Techniques presented herein introduce a framework for finetuning multiple adapters for families of large language models (LLMs).

Active Mini-Tank Immersion Cooling, Stephen A Scearce, Phong Ho, Joel Richard Goergen, Hari Pillai, Rakesh Bhatia

Defensive Publications Series

With increased power density in artificial intelligence (AI) servers, there is increased heat produced. However, the present solutions are not able to remove the heat efficiently or provide for the full capability of the processors, either in a central processing unit or a general processing unit. Presented herein is a hybrid liquid cooling system that allows modular Active Mini-tank immersion cooling (AMTIC) modules as the final stage, in conjunction with or without Direct liquid cooling loops, to facilitate server cooling.