Digital Commons Network^™

System And Method For Ai-Generation And Personalized Music Playback In Social Venue Environments, Eric Zavesky, Miguel A. De La Lama

Defensive Publications Series

A system and method for generating and playing personalized music in a venue setting from multi-modal patron input. The system comprises a multi-modal input interface that accepts musical preferences, narrative text, voice, visual content, and social media links through QR-code-accessed mobile web, kiosk, and voice channels, fusing said inputs into a unified generation request. An AI music generation engine employs a progressive generation pipeline that begins generating preliminary musical content during input collection and refines iteratively as additional input arrives, producing complete songs with lyrics, melody, arrangement, and production. A venue audio integration module queues and plays generated songs through …

Tamper-Evident, Non-Downgradeable Binding Of Advisory Header Metadata To Ciphertext In Client-Side-Encrypted Document Envelopes Via Aes-Gcm Additional Authenticated Data, David Lee Brown

Defensive Publications Series

An encrypted-document envelope carries advisory metadata — a threshold/quorum requirement, a lifetime policy, a distribution policy,   and a lineage pointer — alongside ciphertext. Conventionally this metadata sits in a plaintext or merely hashed header, where a   third party can strip it, alter it, or downgrade the format to disable the behavior it governs, because the metadata is not bound to   the ciphertext itself. An unkeyed header hash does not solve this: an attacker who edits the header simply recomputes the hash.

This disclosure describes binding that metadata to the ciphertext as AES-GCM Additional Authenticated Data (AAD), computed over a   deterministic …

Generation Of Synthetic Training Data Using Randomization In Camera Sensor Response And Image Signal Processing Pipeline, Ahmed Abdelkader, Wen-Sheng Chu

Defensive Publications Series

Successful deployment of machine learning models for computer vision on consumer devices requires generalization across multiple devices with varying camera sensor characteristics and image signal processing (ISP) tuning profiles. To ensure model performance, expensive data collection campaigns need to be executed to capture device-specific data for various devices to obtain training datasets to train such machine learning models. This disclosure describes techniques to generate training data to train a machine learning model for deployment across multiple target devices. Variations in camera sensor and ISP tuning profiles are simulated using randomized RAW transformations to produce different color sensitivities and by randomizing …

Steganography-Based Anti-Abuse Mechanism For Media Provenance, Bryan Jeffery Moles, Brian Jeremiah Murray

Defensive Publications Series

Mechanisms for media provenance are implemented by having a device cryptographically sign a bundle of assertions/claims about the media of interest. These assertions/claims reveal information about the media file, such as whether the media came from the camera of the device, whether it was processed in some way, etc. The trustworthiness of media provenance assertions by such mechanisms is reliant on the signing device being uncompromised. This disclosure describes the use of steganographic techniques to prevent scalable abuse of media signing mechanisms through the use of compromised devices. Per the techniques, the signing mechanism is implemented to inspect media content …

Device Platform-Aware, Closed-Loop Resource Scheduling And Handover Management In A Cellular Network, Fernando Selma, Francis Bourque, Hung Tsang, Jibing Wang, Alex Gorokhov

Defensive Publications Series

Cellular network architectures lack visibility to device-level or application-level states and cross-device awareness. Consequently, network scheduling, resource allocation, and handover are purely reactive and inefficient. This document describes techniques for closed-loop platform-driven resource allocation and proactive mobility management. Per the techniques described herein, a platform coordination server that operates independently of the RAN of the 3GPP cellular network receives UE-level telemetry data that includes application-level demand, location and velocity, and sends an aggregated traffic (or demand) forecast to the network for each cell. The platform coordination server converts a confirmed network resources scheduling plan prepared by a gNB to application …

Non-Disclosing Detection Of Material Evidence Gaps And Hidden Conflicts In Permission-Restricted Enterprise Rag, Niveditha Balakrishna Bhat

Defensive Publications Series

Enterprise retrieval-augmented generation (RAG) systems commonly enforce document-level or content-level permissions before presenting retrieved evidence to a user. Correct enforcement of those permissions, however, does not guarantee that the evidence remaining visible to a particular role is materially sufficient to answer a question safely. A role may retain access to plausible, apparently authoritative evidence while newer rules, material exceptions, or contradictory evidence remain inaccessible. In such cases, the system can produce an incomplete or misleading answer without violating any access-control rule.

This publication describes a role-conditioned evidence-auditing mechanism for detecting such failures before deployment. For a representative question and role, …

Zero-Shot Widget Configuration Via Natural Language Query, Ant Oztaskent, François Deschênes

Defensive Publications Series

Setting up a helpful tool on a phone or computer screen, like a live tile or weather display, usually takes too many steps. This publication describes a new, automated system that lets a user simply type or speak what they want, such as “weather in London,” and the device handles the rest. By acting as a smart broker (e.g., a multi-tier routing system), the system automatically finds the right application, places a live widget, glanceable, or dynamic home screen applet on the screen, and sets up the widget perfectly without any extra tapping from the user. If a matching application …

Dynamically Adjustable Photoplethysmography Geometry Based On Perfusion Index And Motion Profile To Balance Accuracy And Power Efficiency, Hamed Vavadi, Kaiyuan (Kai) Yao, Chintan Trehan

Defensive Publications Series

Wearable photoplethysmography (PPG) sensors detect a small pulsatile signal against a large non-pulsatile background.  The signal quality varies widely across users, activities, and conditions.  This publication describes systems and techniques in which the optical geometry of a PPG sensor is adjusted during operation to balance signal quality against power consumption.  A voltage-controlled film, for instance, over an emitter and a photodetector changes an effective aperture, and thereby the depth of light penetration and the perfusion index, without moving parts.  A control system uses a perfusion index and a motion level to select the aperture size produced by the voltage-controlled film. …

Hierarchical Context Caching Using Data Segmentation And Predictive Aggregation, Sarita A. Joshi, Leo J. Gustas, Blane C. Clark

Defensive Publications Series

Standard context caching for large language models can be inefficient in workflows with large, multi-part contexts, such as clinical documentation, where streaming data can invalidate static records and individual records may not meet caching thresholds. A computer-implemented system and method for optimizing context caching are described. The system may utilize a hierarchical structure that segments input context into static, semi-static, and dynamic layers based on update frequency. The system may also proactively aggregate historical records from multiple patients into a single cache prefix to meet service provider token minimums. This approach of hierarchical segmentation and predictive aggregation can reduce cache …

Warpage Control Of Printed Circuit Boards For Large-Scale Package Assembly Using A Localized Stiffener, N/A

Defensive Publications Series

The technology described in this paper relates to an approach for mitigating the issue of thermal warpage mismatch between high-performance processor packages and system boards by mechanically coupling a localized stiffener structure to a printed circuit board near the mounting area of a large-scale package. By selecting materials with tailored coefficients of thermal expansion, the localized stiffener actively influences the board's thermal warpage profile during solder reflow, forcing it to conform to the warpage shape of the package. This approach significantly reduces assembly defects, such as non-wet opens and bridging, particularly across massive ball grid array footprints. The disclosed technology …

Stiffeners For Package Warpage Control, N/A

Defensive Publications Series

The technology described in this paper relates to a packaging structure and associated methods for reducing out-of-plane deformation. The structure introduces a stiffening element bonded to the bottom side of a package substrate to counteract the coefficient of thermal expansion of semiconductor chips positioned on the top side of the substrate. This arrangement creates a highly symmetrical structure that effectively minimizes warpage without consuming valuable top-side surface area. The bottom-mounted element may feature various openings to accommodate solder balls and other electrical components. By reducing overall package warpage, the structure facilitates the use of finer pitch connections and enhances overall …

Risk Manager Predictive Fraud Futures, Gayathri U, Sitaram Panigrahi

Defensive Publications Series

The present disclosure relates to a system and method for predicting merchant-level fraud spikes in payment transaction networks using Temporal Graph Neural Networks (TGNN). The present disclosure suggests receiving historical payment transaction data associated with a plurality of transactions. The present disclosure also suggests generating, from the historical payment transaction data, a plurality of heterogeneous transaction graphs representing transaction relationships across one or more time intervals. Node feature representations and edge feature representations are generated for the heterogeneous transaction graphs based on the transaction data. Subsequently, the present disclosure suggests determining node-specific temporal context information using transaction timestamps associated with …

Root Cause Investigation Memory For Artificial Intelligence (Ai) Agents, Necati Cehreli

Defensive Publications Series

Artificial intelligence (AI) troubleshooting agents can lose root cause analysis (RCA) investigation state when conversation history is compacted or an investigation is interrupted and resumed. To address the challenges of maintaining RCA investigation state, the techniques presented herein implement a memory system that maintains structured RCA memory outside the chat transcript. The memory tracks diagnostic coverage, unexplored evidence, and candidate root causes; refreshes stale state before compaction; and updates the investigation when new evidence is added. A candidate-cause ledger supports final causal adjudication, reducing the likelihood that a downstream event is mistaken for the underlying root cause.

Novel Crystalline Polymorph Of Janus Kinase Inhibitor, Lee Pharma Limited

Defensive Publications Series

The present disclosure relates to {1-(ethyl sulfonyl)-3-[4-(7H-pyrrolo[2,3-d] pyrimidin-4-yl)-1H-pyrazol-1-yl] azetidin-3-yl} acetonitrile, is represented by following structural formula (1) and its polymorph thereof.

User Authorization Of Ai Generated Code Based On Pre-Execution Telemetry, Timo Hoyer

Defensive Publications Series

Generative artificial intelligence (genAI) technology enables users to specify natural language commands and obtain executable code and/or database queries as AI output. While such code can help users accomplish goals quickly, in enterprise computing environments, it can introduce operational issues/vulnerabilities, waste computational resources, lead to platform failures, and/or administrative friction. This disclosure describes a pre-execution orchestration architecture and dynamic user interface for user authorization of code prior to execution. A user can use genAI to generate code. Prior to execution, such code (executable code, database queries, or infrastructural commands) is subjected to analysis to generate pre-execution telemetry that includes factors …

Forecasting Demand At Physical Facilities Using Diverse Heterogeneous Real-World Data, Kapil Parakh, Daniel Klein

Defensive Publications Series

Unpredictable fluctuations in customer demand at physical facilities can have substantial operational impacts. Current demand forecasting approaches tend to be simplistic and reactive as they rely primarily on internal historical data, which provides inadequate accuracy to support future resource allocation. This disclosure describes techniques to provide accurate medium-term demand forecasts for physical operational facilities by synthesizing a diverse set of relevant real-world factors obtained from a fusion of numerous heterogeneous data sources related to human practices and needs. The factors can be continually aggregated and input to a suitably trained machine learning model trained to generate specific actionable output designed …

System For Orchestrating Multi-Modal Generative Pipelines For State-Consistent Interactive Environments, Parnika Singhal, Chirag Saini, Nilanjana Thakur

Defensive Publications Series

Generative systems for dynamic interactive environments may face technical challenges related to maintaining logical consistency and narrative coherence, sometimes referred to as state drift. A computer-implemented system, such as a server or cloud computing platform, can utilize an orchestration layer to coordinate multiple specialized generative pipelines, for example, for narrative, visuals, and audio. A technique can involve constraining a generative language model to output a structured data object that separates a persistent world state vector from an ephemeral narrative payload. The persistent state vector can be reinjected into the model on subsequent turns to help maintain long-term coherence, while metadata …

Cable Coiling Detection Using Coupling-Based Sensing, Hp Inc

Defensive Publications Series

A smart cable monitoring system detects cable coiling conditions before cable degradation occurs. The system utilizes dedicated sensing conductors integrated within the cable to monitor changes in coupling characteristics caused by cable geometry variations. By comparing the measured sensing signature against a baseline condition, the system can identify prolonged coiled states and provide user feedback or preventive actions. Thermal sensing may also be incorporated to improve detection robustness and support cable health assessment.

Host No-Op Mode, Hp Inc

Defensive Publications Series

There are various attack vectors over USB such as injecting malware, HID spoofing, juice jacking etc. This issue can be mitigated by disabling the USB for good, leaving the functionality over USB unavailable to the user. This problem is more prominent for gadgets/peripherals like chargers/charging cables, storage devices (thumb drives/hard disks), mice/Keyboard etc. connected to the host via USB.

Orientation-Based Eye Safety Control For Floor-Facing Optical Sensors In Mobile Robots, Hp Inc

Defensive Publications Series

An orientation-dependent control system for floor-facing optical modules in a mobile ground vehicle is described. An orientation sensor supplies pitch and roll, or equivalent gravity-vector information, to a controller. If the vehicle remains outside a permitted orientation for a validation period, the controller disables, disconnects, or reduces optical emission. Normal operation is restored after the vehicle returns to a permitted orientation for a recovery period. The arrangement reduces unintended direct exposure during lifting, carrying, servicing, or transport and can reduce power use while the modules are not needed.

Hybrid Reaction Bridge, Hp Inc

Defensive Publications Series

Hybrid meeting platforms give remote attendees structured side-channel mechanisms—virtual hand raises, emoji reactions, poll responses—that are entirely absent for in-room participants, whose non-verbal behaviour remains visible only inside the physical room. This paper describes a bridging system that observes in-room participants through a shared room camera, detects a defined vocabulary of natural non-verbal signals (nodding, head-shaking, raised hands, and postural cues), and translates each qualified signal into a corresponding structured reaction event within the conferencing platform’s existing reaction model. The result is a unified reaction channel in which in-room and remote participants are represented with equivalent expressive bandwidth, without requiring …

Intent-Aware Human Motion Interpretation, Hp Inc

Defensive Publications Series

The central challenge of camera-based gesture input in video meetings is distinguishing deliberate interaction gestures from the natural hand motion that accompanies speech. This paper describes a system that meets that challenge through an intent classification engine evaluating kinematic, temporal, postural, spatial, and speech-aligned features over sliding temporal windows. Qualified gestures pass to a gesture-to-structure transformation pipeline that produces clean, labelled diagram elements—not freeform ink—and renders them as first-class shared content within the conferencing platform, visible to all participants in real time, without requiring specialised hardware.

Interaction Claim Standard, Brian Harber

Defensive Publications Series

Independently implemented systems increasingly produce, exchange, and interpret claims about AI interactions: what participated, which interaction the supporting evidence concerns, and how far an assertion reaches. Where no shared requirements govern these claims, agreement on their meaning across implementations remains incidental rather than normative—each implementation may interpret the same assertion differently.   This document defines the Interaction Claim Standard, a standard for expressing and interpreting claims about AI interactions across independently implemented systems. It standardizes four normative objects: claim semantics, interaction-specific evidentiary binding, claim boundaries, and normative interpretation requirements. Together, these establish cross-implementation agreement on the meaning of claims, their evidentiary …

Integrated Clamp Interface For Joining Mjf Additively Manufactured Components To Non‑Mjf Structural Members, Hp Inc

Defensive Publications Series

This disclosure describes an integrated fastening architecture that enables direct attachment of Multi Jet Fusion (MJF) additively manufactured components to structural elements manufactured from other materials. The solution uses a clamp geometry incorporated into the MJF component, a guided fastening mechanism, and controlled mating faces. The design enables rapid assembly, repeatable retention performance, and non-permanent attachment while minimizing additional hardware. Unlike adhesive bonding, interference fits, or external clamps, the approach separates assembly torque requirements from retention-force generation, improving manufacturing robustness and reducing sensitivity to torque variation. Figure 1a, Figure 1b, and Figure 2 illustrate representative fastening concepts and the screw-held …

Rubber Foot Speaker, Hp Inc

Defensive Publications Series

This disclosure proposed on repositioning the speaker from internal (inside the chassis) to the external surface. The design incorporates the use of rubber foot with speaker into a single component.  The intention of this idea is to reduce the impact of vibration from the speakers and eliminating the rattling issue when installed internally. The idea also promotes sound quality, flexibility and cost reduction.

Server Tray Guide Tool With Anti-Tilt And Winged-Edge Features For Rack Installation, N/A

Defensive Publications Series

Abstract

A guide tool assists with the installation of server trays into data center racks. Constructed from, for example, stainless steel, peek polymer, alumina, and ceramic, the tool is inserted into a rack slot before the server tray to spread the rack’s internal separators. This action creates an unobstructed path, enabling the tray to be seated smoothly and with reduced force. The tool’s design includes winged edges that function as both an insertion stop and a handle for removal. A stepped feature on one end of the tool prevents it from tilting during installation in slim shelf racks. The design …

Expand Iserialnumber In Usb Device Descriptor To Include Configuration Information, Hp Inc

Defensive Publications Series

This disclosure presents a modification to the iSerialNumber field in the USB Device Descriptor to include additional configuration information that manufacturing can use. Printer configuration frequently requires devices to be fully booted up for the manufacturing software to query the configuration.  This approach improves the manufacturing process by providing the needed device configuration at USB device enumeration time.  The manufacturing software can then take the appropriate actions without the need to wait for fully boot up.

Intent-Driven Orchestration For Enterprise Cyber-Physical Systems: Bridging Probabilistic Planning And Deterministic Actuation, Hp Inc

Defensive Publications Series

This disclosure describes an intent-driven orchestration architecture for enterprise cyber-physical systems that bridges generative, probabilistic, or agentic planning and deterministic device actuation. A non-executing, domain-aware planner converts high-level intent into a candidate graph of typed, idempotent Atomic Tool calls, but does not invoke tools or control devices. A deterministic orchestrator executes only graphs that pass four deterministic validation gates covering schema, governance, semantic correctness, and Device Twin-based simulation. Device Twins provide cached state for deterministic simulation, guard evaluation, binding resolution, and observability, but are not consumed by the Planner and are not authoritative for write-path success. Atomic Tools perform semantic …

Domain N Plus X Redundancy And Dynamic Thermal Management For High-Density Compute Domains, N/A

Defensive Publications Series

This paper describes a system architecture for managing high-density compute domains to address thermal and power constraints while facilitating continuous maintenance operations. The system utilizes an N plus x redundancy model for compute units and network switches, where supplemental units are managed as spares in varying low-power states. A central management controller orchestrates power steering and workload migration, aiming to adhere to designated power budgets while avoiding thermal hardware damage. This approach allows for ongoing service availability during hardware repairs and software updates without performance degradation.

Keywords: high-density compute domains, dynamic thermal management, workload migration, power steering, system redundancy, …

Dynamic Exponent-Mantissa Repurposing For Adaptive Microscaling (Demr-Am) Architecture, N/A

Defensive Publications Series

This publication details a hardware architecture and data processing methodology for dynamic exponent and mantissa repurposing, tailored to facilitate adaptive microscaling within machine learning accelerators. The described methodology addresses the quantization errors and memory bandwidth constraints that frequently emerge during the deployment of large-scale artificial intelligence models. By dynamically repurposing redundant exponent bits into additional mantissa bits specifically for outlier values within a given tensor block, the system dramatically enhances numerical precision without expanding the core memory footprint of the data block. A dedicated hardware decoder can be positioned immediately prior to a dot-product engine to re-route electrical signals and …