Digital Commons Network^™

Measuring The Impact Of Network Underlay Outages On Cloud And Production Networks, N/A

Defensive Publications Series

A new approach is provided for measuring the impact of network underlay (Layer 3) outages on customer-facing services, such as internal applications (Layer 7) and external cloud virtual machines (VMs). The approach performs a cross-layer impact analysis by correlating low-level network performance data with customer-level metrics within fine-grained time intervals. By comparing performance across layers for the same network path and time, the system can distinguish between L3 outages that are successfully mitigated ("repaired") and those that "surface" to the customer. The resulting metric, "surfaced SLO burn," quantifies the actual customer-experienced outage, enabling more accurate measurement of the issues and …

Novel And Unobtrusive Technique For Gaze And Head Tracking Calibration In Real-Time For Pc Users, Hp Inc

Defensive Publications Series

This disclosure describes the novel gaze and head tracking calibration procedure for PC users. The process of estimating a participant's ocular and head-pose features to serve as the foundation for a completely personalized and precise gaze point computation is known as calibration. The technique for calibration proposed in this disclosure applies to both types of trackers: external cameras mounted to the PC and software that uses the PC’s built-in camera. The current techniques for such calibrations require the PC user to focus on the pre-defined gaze points at the corners of the screen before gaze-tracking can be performed by software. …

Service Activation Test Solution Over Evpn Multi Homed Networks, Anonymous

Defensive Publications Series

This disclosure relates to enhancements in Service Activation Test (SAT) methodologies for Ethernet VPN (EVPN) networks employing multi-homing with all-active redundancy. Conventional SAT methods, such as those defined in RFC 2544 and ITU-T Y.1564, assume symmetric traffic paths and do not account for Equal-Cost Multi-Path (ECMP) forwarding behavior in EVPN VPLS, VPWS, or FXC deployments. As a result, test frames may traverse non-deterministic paths, causing inaccurate or incomplete service-level measurements. The disclosed solution introduces a uni-directional SAT framework that computes throughput, latency, jitter, and frame-loss on forward paths only, avoiding reliance on reverse traffic. The approach utilizes three functional elements: …

Novel Process For The Preparation Of (R)-3-(4-(7h-Pyrrolo[2,3­D]Pyrimidin-4-Yl)-1h-Pyrazol-1-Yl)-3-Cyclopentylpropanenitrile Or A Salt Thereof, Msn Laboratories Private Limited, R&D Center; Srinivasan Thirumalai Rajan, M Ramakrishna, Chandrakanth Rao Sirigiri And Talasani Kiran Reddy

Defensive Publications Series

Abstract:

The present invention relates to a novel process for the preparation of (R)-3-(4-(7H-pyrrolo[2,3­d]pyrimidin-4-yl)-1H-pyrazol-1-yl)-3-cyclopentylpropanenitrile or a salt thereof, which is referred to as Ruxolitinib and represented by the following structural formulae,

Formula-1

Unsupervised Clustering With A Large Language Model Using Controlled Generation, Leonid Kuligin, Max Tschochohei

Defensive Publications Series

This disclosure describes techniques that leverage the analysis, reasoning, and classification abilities of large language models (LLMs) to cluster text. An initial cluster is formed, and it is represented by a text description (rather than a conventional vector in a high-dimensional space). Guided by controlled generation, the LLM operates in a loop to classify successive input text into new or existing clusters. Each text from the input dataset is assigned to an existing cluster if its description is similar to that of one of the existing clusters. The description of an existing cluster can be merged with the new text …

A Process For The Preparation Of Diazoxide Choline, Srinivasan Thirumalai Rajan, Ghojala Venkat Reddy, Gade Srinivas Reddy, Uppalanchi Srinivas, Manda Ananth Reddy, Boddu Ananda Rao, Gade Srinivas Reddy, Thippireddy Purna Chandrasekhar Reddy And Gadi Kondala Ganesh Kumar Msn Laboratories Private Limited, R&D Center, Hyderabad

Defensive Publications Series

The present application relates to a process for the preparation of Diazoxide choline of formula-1, represented by the structural formula-1 below.

Formula-1.

Fully Deployable Rigid Wing, Corentin Barut

Defensive Publications Series

This disclosure presents a novel fully deployable rigid wing system designed for aeronautical applications, utilizing an innovative scissor mechanism for complete deployment and retraction. The wing, composed of multiple rigid segments split along the chord, employs a dual-motion approach involving translation towards the aircraft's center and rotation from a horizontal to a vertical position for retraction. The system ensures structural rigidity through a primary scissor mechanism, maintained by a minimal holding force, and a secondary scissor set with bars of slightly different lengths to enhance angular range. Deployment and retraction are efficiently controlled by manipulating the initial scissor pair, offering …

Process For The Preparation Of Crystalline Form-A Of 6-(Cyclopropanecarbonylamido)-4-[2-Methoxy-3- (1-Methyl-1,2,4-Triazol-3-Yl) Anilino] -N-(Trideuteriomethyl)Pyridazine-3-Carboxamide, Msn Laboratories Private Limited, R&D Center; Srinivasan Thirumalai Rajan; Revu Satyanarayana; Challa Suresh; Sunkari Suresh Kumar

Defensive Publications Series

The present invention relates to a process for the preparation of crystalline form-A of 6-(cyclopropanecarbonylamido)-4-[2-methoxy-3-(1-methyl-1,2,4-triazol-3-yl)anilino]-N-(trideuteriomethyl)pyridazine-3-carboxamide.

Ai-Driven Generation Of Personalized Movies With Dynamic Character Integration, Gaurav Umeshkumar Gupta

Defensive Publications Series

This disclosure describes a platform that leverages AI to generate personalized cinematic narratives, shifting the user experience from passive viewing to active co-creation. The platform enables users to co-create scripts, integrate their likeness into characters, and experience unique, on-demand movies. The platform operates in two primary modes: a full-generation mode for creating original movies based on user prompts and likeness data, and a variable character movie (VCM) mode where users can insert their likeness into pre-defined variable character slots within professionally produced films. A modular architecture is described that includes an AI orchestration engine, script generation, character integration, video synthesis, …

Amorphous Form Of Zongertinib, Processes For Preparation, And Pharmaceutical Compositions Thereof, Prashanth Venukathala, Prashanth Venukathala

Defensive Publications Series

The present invention relates to solid-state forms of small-molecule kinase inhibitors. In particular, it relates to an amorphous form of Zongertinib (hereinafter “the compound” or “Zongertinib”), processes for its preparation, compositions comprising the amorphous form, and their use in therapy.

Amorphous Brensocatib, Prashanth Venukathala, Prashanth Venukathala

Defensive Publications Series

The present invention relates to the amorphous form of Brensocatib, pharmaceutical compositions comprising the same, and processes for its preparation. More particularly, the invention provides an amorphous preparation that demonstrates improved solubility, dissolution rate, and bioavailability compared to the crystalline form of Brensocatib.

An Improved Process For The Preparation Of Indole Derivative Of Formula (I), Anonymous

Defensive Publications Series

The present disclosure relates to a process for the preparation of indole derivative, a compound of Formula (I) comprising;

Automated Bug Tracking And Resolution Using An Artificial Intelligence Agent, Alankrit Kharbanda, Sowmya Dayanand

Defensive Publications Series

Traditional bug triage and resolution can be inefficient and time consuming. This disclosure describes techniques of bug triage that use artificial intelligence (AI) agents trained on domain-specific data to deliver accurate, actionable insights for bug analysis. A bug triaging AI agent is trained on a dataset that includes historical bug dumps from bugs and associated playbooks, design documents, customer communications over email or messaging, etc. Such targeted training provides domain-specific expertise, enhanced AI agent accuracy (minimizing hallucinations) when generating actionable insights such as potential next steps, identifying potential root causes and relevant internal documentation, providing call-the-expert suggestions, etc. The AI …

Seamless Playback Resumption For Musical Containers, John Oberbeck, Alan Bettes, Chunya Hua

Defensive Publications Series

Music streaming platforms do not typically provide the ability for users to resume listening to musical containers (albums, playlists, etc.) from the point where they last left off. This disclosure describes techniques that enable users to resume listening to musical containers, thereby solving the problem of continuity for music. The techniques enable users to browse older queues not currently being listened to and to resume playback across devices on finite containers while respecting a shuffled order. Furthermore, the techniques provide a visual user interface (UI) to identify a finite container or queue as being in-progress; a visual UI on the …

Multi-Device Asynchronous Distributed Compute Framework Of Llm Layers For Cascaded Inference, Na

Defensive Publications Series

Large language models (LLMs) operating in a server-client framework can incur high latency, particularly for complex queries. This disclosure describes a framework to serially distribute LLM inference layers between compute nodes based on just-in-time query optimization. The division of the transformer layers is parametrized using the number of compute nodes and the frequencies of the queries to enable optimal asynchronous operation for LLM task completion. Due to the asynchronous nature of the described distributed computing and inference, latent and direct queries do not compete with each other, thereby reducing the latency for LLM task completion.

Auto-Spatializing Environments, Dongeek Shin, Nooshin Mojab

Defensive Publications Series

A three-dimensional (3D) experience, in the context of virtual or augmented reality, seeks to replicate the way humans perceive the real world. A true 3D experience typically involves a complete 3D model of the environment, where every object has defined geometry, texture, and position in space. However, creating these fully-realized 3D environments is resource-intensive, requiring significant computational power for real-time rendering and complex algorithms for asset creation. This can be a major bottleneck for applications that need to generate a wide variety of immersive backdrops quickly.

An alternative to the 3D experience is a "3D-like" experience, often referred to as …

Camera-Based Virtual Input, Weichung Chang

Defensive Publications Series

Conventional input methods for compact electronic devices, such as on-screen virtual keyboards, suffer from limitations including cramped layouts, a lack of tactile feedback, and unsuitability for touchless environments like augmented reality. These issues often lead to frequent typing errors and reduced usability for complex tasks. To address these and other challenges, a camera-based virtual input method is described. Using the camera of a device, images of a user's hand and finger movements may be captured as they tap on any surface or imitate typing in the air. A prediction model analyzes these movements, not as absolute positions, but as relative …

An Information-Algebra Framework For An Ai-Driven User Interface, D Shin

Defensive Publications Series

Large language models (LLMs) can be accessed and interacted with via a multimodal chat interface. LLMs can enhance user productivity and information organization. This disclosure describes techniques that extend the interface of the LLM to a beyond-the-chat, free-form interface. The interface utilizes an information algebra framework for LLMs on structured input and output such as sticky notes, symbols, tags, etc. using custom fine-tuned routing models to determine user intent. A suitably intuitive meaning is assigned to ‘add,’ ‘subtract,’ and similar algebra-like operations on structured input and output. As the model matures and reads user intent more accurately, human-AI interactions are …

A System, Method, And Computer Program Product For Dynamically Adjusting Currencies Using Machine-Learning Models, Libby A. Kurien, Joshua Quiroz

Defensive Publications Series

This paper proposes a novel approach of using real time sentiment analysis based on retrieval augmented generation (RAG) for foreign exchange (forex) pricing. It argues that large language models (LLMs) can capture the semantic and emotional aspects of news articles and social media posts that affect the forex market dynamics. Traditional forex trading relies heavily on technical and fundamental analysis. However, the advent of big data and sophisticated text analysis techniques using LLMs have opened up new avenues for analyzing market sentiments in real time that can be extended to Forex. LLMs when used in combination with RAG, an architectural …

Intercept & Solve For Printer Offline, Hp Inc

Defensive Publications Series

This disclosure pertains to the domain of network connectivity, with a focus on methods for identifying the causes of printer offline issues and assisting customers by delivering troubleshooting support through a chat-bot.

Hierarchical Path-Based Signal Generation For Content Recommendations, Tamojit Chatterjee, Shravan Nayak

Defensive Publications Series

Generating recommendation signals for media content items that have good coverage and that include details about the media content item may be challenging, especially for a large media content universe. In some cases, sparse signals may fail to capture nuanced relationships between different media content items, leading to less effective media content recommendations. Existing methods for generating more abstract signals may involve significant manual effort and/or complex data extraction processes.

The disclosed technology for an improved recommendation system addresses these limitations by using hierarchical clustering of media content items (e.g., movies, television shows, short form videos, etc.). The recommendation system …

Using Off-The-Shelf Llm Prompting For Query And Subquery Intent Quality Evaluation, Pulkit Jain, Josh Ain, Vy Thuy Nguyen, Lin Dong, John Cockrell, Aditya Mone, Tushar Chugh

Defensive Publications Series

Evaluation of search query and subquery quality can be beneficial for information retrieval systems, as some evaluation methods that use human raters may be time-consuming, costly, and difficult to scale. This disclosure describes a method and system for automating query quality evaluation that may use a prompted, off-the-shelf large language model (LLM). A process can begin by creating a small, curated "golden dataset" of positive and negative examples, for instance, from historical human ratings. This dataset can then be used to engineer a detailed prompt to guide an LLM to perform classification tasks on new query-subquery pairs, such as identifying …

Self-Learning, Iterative Agent For Root Cause Analysis Of Service Failures, N/A

Defensive Publications Series

A learning agent designed to automate the root cause analysis (RCA) of failures within complex distributed systems. The agent can be configured to perform iterative generation and investigation of hypotheses. When a service failure alert is triggered, the agent may ingest data from various sources, including live monitoring data and historical incident records. Based on this information, the agent may be configured to formulate a ranked list of potential hypotheses for the cause of the failure. The agent then may be configured to systematically investigate each hypothesis by gathering and analyzing evidence from relevant data sources until a root cause …

Tariff Forecasting To Inform Dynamic Pricing And Advertising Bidding For Online Retail, Kshitij Gaikar, Pushkar Mittal

Defensive Publications Series

A system is described for addressing challenges related to international trade tariffs in e-commerce. The system is configured to ingest global tariff schedules, including published future changes, and can use an artificial intelligence (AI) model to map retailer product stock-keeping units (SKUs) to Harmonized System (HS) codes to assist in determining applicable tariffs. The system may then calculate current and anticipated landed costs. This analysis can be used to generate pricing and advertising bid recommendations for retailers and to provide proactive alerts to consumers about potential upcoming tariff-driven price adjustments. Such features may support price and advertising efficiency for retailers …

A System For Disambiguating Code-Switched Speech Using Visual Context, Mithun Kumar S R

Defensive Publications Series

Natural Language Understanding systems can encounter challenges when resolving lexical ambiguity in code-switched speech, particularly with inter-language homonyms that may be phonetically similar or identical but have different meanings. These challenges can be compounded by a reliance on audio-only input, which may not capture environmental context that can be beneficial for disambiguation. The disclosed method addresses this by concurrently processing an audio stream containing a user's utterance and a video stream capturing the user's visual context. The system can be configured to identify potential homonyms in the transcribed speech and to extract semantic tags, such as for objects or scenes, …

Managing Device States Based On Notification And User Context, Harshad Dhabu, Abhijit Adsule

Defensive Publications Series

Computing devices, such as wearable devices, may experience increased power consumption due to device wake-ups associated with high-priority notifications, and some notification management systems can offer limited granular control, for example, by muting notifications on a per-application basis. This disclosure describes a system for context-aware notification handling where alerts may be processed as silent by default. A decision-making engine can analyze a user's context, such as current activity and location, and a notification's context, including its content, sender, and origin application. By combining these factors, the system can determine whether to trigger a device wake-up for a notification that meets …

Container-Based Management Of Ephemeral Applications, Harshad Dhabu, Abhijit Adsule

Defensive Publications Series

The persistent installation of infrequently used applications on a computing device, such as a smartphone or wearable device, may consume storage and processing resources, and some management techniques may not fully address related issues like visual clutter in an application launcher. A method is described for managing applications through a container system that can facilitate on-demand, ephemeral execution. When a user launches an application, a system may dynamically provision an available version from a source, such as a remote application repository, and execute it within an isolated container. Upon session completion or following a period of inactivity, the system can …

Coordinated Data Collection For Event Correlation Analysis, Theodore Obbard, David Phillips

Defensive Publications Series

Independent random sampling of different data types in large-scale applications can make it difficult to analyze correlations between separate events because the data points are often collected independently. The disclosed method provides a system for coordinated data sampling that can be managed by a client-side library. Designated 'data provider' events, such as a page load, may trigger a temporary 'coordinated mode' based on a configurable sample rate. When this mode is active, multiple, distinct 'data consumer' points across the application lifecycle can be instructed to log their granular data for a given session. When not in coordinated mode, these data …

Decomposition Of Website Interactions Into Executable Semantic Functions, Emrick Donadei, Kalyan Desineni, Alan Pham, Kaushik Sethuraman

Defensive Publications Series

Automated agents can experience inefficiency and brittleness when performing tasks on websites, as some user interface (UI) automation methods can be slow or fragile, and public APIs may be incomplete or unavailable. This disclosure describes a method for decomposing website interactions into executable semantic functions through an offline process. The method can involve tracing a user task to collect interaction data, such as UI events, network traffic, and JavaScript execution details. A large language model can analyze this trace to identify logic and components associated with the task. From this analysis, a self-contained, executable function can be generated. This process …

Systems And Methods For Translating Mixed-Language Audio, Mithun Kumar S R

Defensive Publications Series

Some real-time speech translation systems are architected for monolingual input and can encounter challenges, such as transcription errors or increased latency, when processing code-switched speech. The disclosed method describes a system for real-time interpretation that can ingest a raw audio stream and utilize a dual-axis model to concurrently identify both a spoken word and its language of origin. This process can produce a token stream with associated language identifiers. A subsequent syntactic bridge module can analyze this mixed-language stream to generate a language-independent grammatical representation of the utterance's semantic content. This abstract structure may then be used to synthesize a …