Open Access. Powered by Scholars. Published by Universities.®
- Publication Year
- File Type
Articles 4771 - 4800 of 11865
Full-Text Articles in Entire DC Network
Striped Mamba: A Novel Approach To Parallel Processing In State Space Models, Renee Gagnon
Striped Mamba: A Novel Approach To Parallel Processing In State Space Models, Renee Gagnon
Defensive Publications Series
This document discloses a novel technique called "Striped Mamba" for enhancing the parallelism and efficiency of State Space Models (SSMs) in neural network architectures. The Striped Mamba technique introduces a method of dividing input sequences into multiple "stripes" that can be processed in parallel, while maintaining the sequential dependencies crucial to SSMs. This approach aims to leverage the parallel processing capabilities of modern GPUs while preserving the temporal modeling strengths of SSMs.
Caching Of Retrieval Augmented Generation Model Trained With Dynamic Data And Dynamically Updating Of Cache, Mustafa Kadioglu, Paul C. Stojanovski, Sunil Karthik Kota
Caching Of Retrieval Augmented Generation Model Trained With Dynamic Data And Dynamically Updating Of Cache, Mustafa Kadioglu, Paul C. Stojanovski, Sunil Karthik Kota
Defensive Publications Series
Techniques are presented herein provide for an intelligent caching mechanism of a Retrieval Augmented Generation (RAG) model trained with dynamic data which is subject to get deleted or modified periodically. Based on the feedback from the customer, Cache Vector DB is updated. This method is capable of processing positive and negative feedback from customers to update the Cache Vector DB while simultaneously protecting from negative feedback spam. The method leverages a framework that not only caches an original question/query (primary) and the corresponding response, but also secondary question/queries which either uses GenAI to create synthetic queries of the original question/query …
Power Efficient Connectivity Between Mobile Devices, Nazih Almalki, Andy Killen
Power Efficient Connectivity Between Mobile Devices, Nazih Almalki, Andy Killen
Defensive Publications Series
Mobile devices, such as wearable computing devices (e.g., smart watches), may default to a particular radio for tethering (e.g., connectivity) with a computing device (e.g., a phone). In accordance with the techniques disclosed herein, a wearable computing device may utilize a connectivity context to proactively provide power efficient connectivity between the wearable computing devices and the computing device. The wearable computing device may select a radio from a plurality of onboard radios based on the connectivity context. The connectivity context may indicate conditions for which the selected radio may be better suited for connectivity (e.g., data transfer) with the computing …
System And Method For Configurable Vehicle Metrics Collection And Reporting, Selim Gurun, Umang Bhatia, Rui Qiu, Sergey Grizan
System And Method For Configurable Vehicle Metrics Collection And Reporting, Selim Gurun, Umang Bhatia, Rui Qiu, Sergey Grizan
Defensive Publications Series
This publication describes systems and techniques for implementing a vehicle communications language for describing vehicle system data sources, data collection rules, data processing, and data results using Protobuf for serializing data and establishing an interface description language for communications among various vehicle systems and vehicle data sources. Described techniques may enable vehicle data collection, vehicle data configuration and report generation. Aspects of the disclosure may include end-to-end cloud integration for vehicle systems and data sources utilizing a digital twin of a subject vehicle within the infrastructure of a cloud services provider and defining within the digital twin, available vehicle services, …
Provides Convenience Of App Installation Through App And Printer Management Using Tags, Hp Inc
Provides Convenience Of App Installation Through App And Printer Management Using Tags, Hp Inc
Defensive Publications Series
In the legacy enterprise solution, when users create policy, users had to select devices or application one by one to change settings or install application. And If user subscribe new application, user has to modify or create policy for new application. To solve this inconvenience this idea provide TAG combination.
Auto Generation Of User-Defined Print Option Based On Document Analysis, Hp Inc
Auto Generation Of User-Defined Print Option Based On Document Analysis, Hp Inc
Defensive Publications Series
Printer driver supports printing features such as Watermark, Special Page, and Tab Printing features, which users can select options from multiple sub-features in each feature's settings window and create a new single option from the combination of these selections. This article is suggested to provide Auto generation option in printer driver UI for users in order to improve usability of setting options and provide the enhanced print output. To print with selecting the Auto generation option analyzes the contents of the document and automatically applies the printing features to generate print output preview. When users agree with the automatically generated …
Eddy Current Non-Destructive Evaluation Of Stator Winding Weld Quality, Anonymous
Eddy Current Non-Destructive Evaluation Of Stator Winding Weld Quality, Anonymous
Defensive Publications Series
No abstract provided.
Zebralstm: An Advanced Lstm Architecture For Efficient Sequential Data Processing, Renee Gagnon
Zebralstm: An Advanced Lstm Architecture For Efficient Sequential Data Processing, Renee Gagnon
Defensive Publications Series
ZebraLSTM is a novel Long Short-Term Memory (LSTM) architecture designed to optimize the performance of sequential data processing tasks. By dividing the hidden state into multiple parallel stripes, ZebraLSTM enhances computational efficiency and scalability, particularly when leveraging modern GPU architectures. This document outlines the design, implementation, and potential applications of ZebraLSTM in various domains, including natural language processing, time-series analysis, and drug design.
Large Language Model Query Routing On A Controller, Eric A Voit, Samer M Salam, Pradeep K Kathail, Arman Rezaee
Large Language Model Query Routing On A Controller, Eric A Voit, Samer M Salam, Pradeep K Kathail, Arman Rezaee
Defensive Publications Series
A system that allows authoritative ownership of documentation for a particular product and organizational role is proposed herein. The system may allow a controller or other to system to not attempt to provide chatbot answers when there are no relevant retrieval augmented generation (RAG) retrieved documents for the particular product, even when a similar product might support a given capability queried for the particular product.
Pressure Equalization For Liquid Co2 Pipe Segments With Closed/Blocked Outlet
Pressure Equalization For Liquid Co2 Pipe Segments With Closed/Blocked Outlet
Defensive Publications Series
Pressure equalization for Liquid CO2 pipe segments with closed/blocked outlet through bypass valves which route back to the storage tank
Smart Lock Battery Charging Via Wired Smart Doorbell, David Wang, Jimmy Lim, Arun Rao
Smart Lock Battery Charging Via Wired Smart Doorbell, David Wang, Jimmy Lim, Arun Rao
Defensive Publications Series
Current smart locks have several drawbacks - the smart locks are large and thick, and have an unappealing visual appearance; need periodic battery replacement; are inoperable when the battery is being replaced and can cause users to be locked out; have an environmental impact due to large battery; and can suffer from battery performance in low temperature conditions, which reduces the usable runtime. This disclosure describes a smart doorbell and smart lock powered and charged by a home doorbell transformer. The power path to the smart lock is through a smart doorbell that includes a controller. The smart lock has …
Creation/Extension Of Software-Defined Fabric Using Networking Llms With Augmented Learning, Prakash C Jain, Sanjay K Hooda
Creation/Extension Of Software-Defined Fabric Using Networking Llms With Augmented Learning, Prakash C Jain, Sanjay K Hooda
Defensive Publications Series
Large Language Models (LLMs) can be used in a variety of environments to enhance user experience/understanding of information related to different scenarios. However, in the context of software-defined fabric management and/or design tasks, an LLM may provide responses to fabric management/design queries that may be of limited use to a network operator. Proposed herein are techniques for augmenting dynamic runtime information (e.g., topology, configuration, events, etc.) of a software-defined fabric to an LLM using a vector database to enable the LLM to provide recommendations to a user/administrator of the fabric that are based on the context as opposed to providing …
Pictorial Key-Pad, Colin Hilton
Pictorial Key-Pad, Colin Hilton
Defensive Publications Series
Despite advances in biometric access to smartphones and tablets, passcodes are set to remain in use due to their simplicity and accessibility. Means of inputting passcodes using numerical keypads has proven to be vulnerable, however, to fraudulent access. The methods here leverage the increased size and resolution of screens to implement a more secure way to encode both personal identification numbers and passcodes.
Ai-Powered Personalized Output Settings(Eco-Mode) Suggestions To Save Customers Cpp(Cost Per Page), Hp Inc
Ai-Powered Personalized Output Settings(Eco-Mode) Suggestions To Save Customers Cpp(Cost Per Page), Hp Inc
Defensive Publications Series
Typically, the yields shown when you purchase toner are based on standard charts, so when you actually print a document, you might experience less yield than shown. You'll also use the toner eco mode provided by the manufacturer to increase yield. Although the savings for eco mode is shown, the savings shown may vary depending on the documents you are using, as the various algorithms for eco mode are based on the actual content of the documents you are using. Therefore, the information presented about consumable usage and features to save consumables is generalized and may not apply to your …
Correlation Of Extended Berkeley Packet Filter Data, Network Flow, And Application Data To Achieve Deeper Insights, Tim Szigeti, Dave Zacks, Ted Hulick
Correlation Of Extended Berkeley Packet Filter Data, Network Flow, And Application Data To Achieve Deeper Insights, Tim Szigeti, Dave Zacks, Ted Hulick
Defensive Publications Series
Proposed herein are techniques to correlate together data derived from extended Berkeley Packet Filter (eBPF) visibility on hosts, combined with network data, such as that derived from monitored network traffic, and end host data, such as that derived from network visibility tools or other similar network monitoring tools/sources, in order to drive deeper visibility into traffic flows across network, and potential policy outcomes in mid-span network devices based on these correlations and the insights they provide.
Multi-Tenant Least Material Carbon Path Selection, Chih-Tsung Huang, Michael King, Ajay Ranjith Vempati, Wei-Jen Huang
Multi-Tenant Least Material Carbon Path Selection, Chih-Tsung Huang, Michael King, Ajay Ranjith Vempati, Wei-Jen Huang
Defensive Publications Series
Sustainability has become a very important topic, particularly in the area of networks/network products. In particular, significant industry investments are being made to collect full material disclosure (e.g., a full material breakdown of a product, system, etc.) and companies are developing innovations to monetize and incentivize greener products. This proposal provides techniques through which a least Material Carbon Footprint (MCF) path can be determined for a network based on various material disclosure information that can be gathered/determined for one or more products of the network.
Multimodevote: Multimodal Poll Participation System, Peter Hraška, Dávid Domonkos, Marek Šuppa, Ondrej Močkor, Miloš Lajtman
Multimodevote: Multimodal Poll Participation System, Peter Hraška, Dávid Domonkos, Marek Šuppa, Ondrej Močkor, Miloš Lajtman
Defensive Publications Series
Polls are essential for gathering feedback and making decisions in virtual meetings. However, conventional voting methods often lack immediacy and restrict users to clicking buttons. Proposed herein is a system that provides the possibilities of poll participation by allowing users to cast votes using emoji reactions, facial expressions or gestures in virtual meetings, message apps and other scenarios.
Geospatial Intelligence Squad, Sumit Kumar, Visa, Mallikarjun Ambati, Visa, Monish Kumar Sinha, Visa, Anoop Dubey, Visa, Anna Bedrosova, Visa
Geospatial Intelligence Squad, Sumit Kumar, Visa, Mallikarjun Ambati, Visa, Monish Kumar Sinha, Visa, Anoop Dubey, Visa, Anna Bedrosova, Visa
Defensive Publications Series
The present disclosure relates to techniques for empowering geospatial data interaction and map visualization with Natural Language Processing (NLP) via Generative Pre-trained Transformers (GPT) Application Programming interfaces (APIs) using a system. The proposed system may allow the user to input the query using user interface. The input in transmitted or sent to the server through a communication network. The server may send the input to the GEN AI model. The GEN AI model may process the input and send the response to the server. Subsequently, the server send the response to the UI. Finally, the output is provided to the …
Personalized Navigational Guidance With Context-Based Lead Time, Jonathan C. Chu, Daniel V. Klein, Paul Mccartney
Personalized Navigational Guidance With Context-Based Lead Time, Jonathan C. Chu, Daniel V. Klein, Paul Mccartney
Defensive Publications Series
Digital map and navigation applications provide dynamic turn-by-turn guidance. However, such guidance does not take into account a user’s prior knowledge of portions of a route and may also fail to take into account real-time conditions on the route. Thus, a user may receive unnecessary instructions for familiar road segments and may not receive guidance with sufficient lead time, causing rerouting and delays. This disclosure describes techniques, implemented with user permission to access relevant user data, to provide personalized live navigational guidance with detailed directions delivered only during unfamiliar portions of a route. The instructions are personalized based on a …
Statistical Techniques Of Regression Testing To Validate The Correctness Of New App Editions, Grant Kennell, Fei Wang, Christopher Charles Burns, Zening Li, Hsin-Pei Lee, Jessica Hua, Sam Cunningham, Shu-Wei Cheng, Eugene Krasichkov, Lee Williams
Statistical Techniques Of Regression Testing To Validate The Correctness Of New App Editions, Grant Kennell, Fei Wang, Christopher Charles Burns, Zening Li, Hsin-Pei Lee, Jessica Hua, Sam Cunningham, Shu-Wei Cheng, Eugene Krasichkov, Lee Williams
Defensive Publications Series
Regression tests are conducted on new editions of an app to ensure backward compatibility with previous editions. Due to testing complexity, live, end-to-end regression tests are relatively rare. Therefore, regression tests may not truly reflect software behavior from a user perspective. This disclosure describes the use of statistical significance techniques to regression-test new editions of an app. Two editions are given similar inputs and configurations, and the user-facing outputs, e.g., screenshots of viewports of the user interface, are subjected to statistical significance tests that output a likelihood that the two outputs arise from the same source. Insubstantial differences are not …
Rapid Video Prototyping And Content Creation Using Generative Artificial Intelligence, Krishna Chytanya Ayyagari
Rapid Video Prototyping And Content Creation Using Generative Artificial Intelligence, Krishna Chytanya Ayyagari
Defensive Publications Series
Traditional video production is a resource intensive process, requiring significant time and large financial investment. A common practice in entertainment and gaming industries is to generate prototypes or previews of video content before final video creation. This disclosure describes the use of generative artificial intelligence to automatically generate short video content that can be used directly as short-form video content (e.g., advertisements, short-form videos, etc.) or can be used as part of a larger video that may also include human-created content. Video generation requests are received as prompts, which are refined and processed to ensure compliance. Generative video libraries are …
Detection Of Anomalies In Visual Elements, Shu-Wei Cheng, Grant Kennell, Fei Wang
Detection Of Anomalies In Visual Elements, Shu-Wei Cheng, Grant Kennell, Fei Wang
Defensive Publications Series
As new versions of an app or other software are checked into a version control system, bugs can be introduced that are undetected by regression testing. Apps or software that have visual elements (VE) or components are particularly susceptible because regression tests usually do not include comparisons of viewports as seen by a user. This disclosure describes techniques to detect bugs in successive versions of a software under test by translating visual elements into quantitative metrics; by detecting anomalies, e.g., sudden spikes, cliffs, or drops, in quantitative metrics; and by mapping the detected anomaly to particular versions of constituent module(s) …
Bfd/S-Bfd Discriminator Collision Detection And Recovery, Anonymous
Bfd/S-Bfd Discriminator Collision Detection And Recovery, Anonymous
Defensive Publications Series
The present disclosure relates to Bidirectional Forwarding Detection (BFD) and Seamless BFD (S-BFD), namely discriminator collision detection and recovery. Specifically, a new BFD diagnostic code is presented, e.g., “BFD discriminator cross-connect (xcon).” A BFD session is held in the down state when “BFD discriminator xcon” is detected. On detection on “BFD discriminator xcon,” a new local discriminator value be chosen for session on which fault is detected. This prevents undesirable flapping of BFD session providing better stability of network. With an alarm for discriminator collision, early corrective actions can be taken to recover from such conditions.
Providing Llm-Generated Point Of Interest Description Based On Gaze Tracking, Dmitrii Boiarshinov, Jaime Guajardo, Gabi Lanning
Providing Llm-Generated Point Of Interest Description Based On Gaze Tracking, Dmitrii Boiarshinov, Jaime Guajardo, Gabi Lanning
Defensive Publications Series
Visually impaired individuals face challenges in understanding the spatial location of objects around them, particularly for objects that are at a distance. The lack of spatial awareness can lead to difficulties in navigation and interaction with the environment. This disclosure describes the use of an augmented reality (AR) headset and a multimodal large language model (LLM) to automatically provide useful scene descriptions to users. The headset includes world-side cameras that capture visual information in a user’s surroundings. The headset also includes eye-side cameras that, with user permission, detect the direction and focus of the user’s gaze. The detected eye movements …
Advanced Office Experience Through Networked Docks, Hp Inc
Advanced Office Experience Through Networked Docks, Hp Inc
Defensive Publications Series
The modern office environment has evolved. Similar concepts such as hot desks, mobile seats, flex desks emerge and now hybrid work became the new norm. In this dynamic setting, providing employees with seamless connectivity and enhancing multi-tasking abilities are essential. Traditionally, docking stations have been utilized to connect external monitors, peripherals and to provide more I/O ports extension. Furthermore, since dock locations are fixed and evenly distributed across office sections based on floor plans, we believe they can serve as an effective tool for advancing office management and providing a more comprehensive office experience. This idea describes an innovative office …
Creating Virtual Meeting Spaces Based On In-Person Discussions Using Location Services Platforms, Deepesh Arora, Rajarshee Dhar, Ram Mohan R, Gonzalo A Salgueiro
Creating Virtual Meeting Spaces Based On In-Person Discussions Using Location Services Platforms, Deepesh Arora, Rajarshee Dhar, Ram Mohan R, Gonzalo A Salgueiro
Defensive Publications Series
Techniques described herein provide a mechanism for in-person discussions to be broadcasted as “virtual spaces” so that relevant users can view the virtual spaces and join the discussions. The relevant users may request to join the virtual space and may be approved or denied. The techniques described herein provide easier collaboration opportunities in a hybrid meeting ecosystem.
Low Power Indoor Standard Power Relay, Domenico Ficara, Pascal Thubert, Jerome Henry, Brian Hart, Amine Choukir, Sachin D Wakudkar
Low Power Indoor Standard Power Relay, Domenico Ficara, Pascal Thubert, Jerome Henry, Brian Hart, Amine Choukir, Sachin D Wakudkar
Defensive Publications Series
Based on the equipment classes for Part 15 Subpart E Unlicensed National Information Infrastructure (U-NII) defined for 6 GHz frequency band U-NII device certifications, low power indoor (LPI) only clients cannot connect to standard power (SP) access point. To overcome this issue, techniques are proposed herein that provide for the LPI only clients to reach the SP access point by means of relaying nodes.
On-Pc Ai-Powered Malware Detection Via Screen Buffer, Hp Inc
On-Pc Ai-Powered Malware Detection Via Screen Buffer, Hp Inc
Defensive Publications Series
This publication dives into the innovative use of on-PC artificial intelligence (AI) hardware to enhance malware detection and prevention. Specifically, it focuses on detecting and blocking malicious imitation of an Operating System (i.e.: Microsoft® Windows®*) login user interface (UI) in real-time through AI-powered screen buffer analysis.
Custom Trust Attribute For Iot And Iiot Devices In Manufacturer Usage Description (Mud) Profile To Mitigate Advance Persistent Threats (Apts), Sindhuja Rao, Raghunath Ramesh Kulkarni
Custom Trust Attribute For Iot And Iiot Devices In Manufacturer Usage Description (Mud) Profile To Mitigate Advance Persistent Threats (Apts), Sindhuja Rao, Raghunath Ramesh Kulkarni
Defensive Publications Series
A custom trust attribute is for Internet of Things (IoT) and Industrial Internet of Things (IIoT) devices in a Manufacture Usage Description (MUD) profile is provided to mitigate Advance Persistent Threats as part of a dynamic security model enhancement for IoT networks. A trust attribute called "trust-status" is defined in the MUD profile. This can be used to continuously adjust IoT device trust levels based on real-time behavior, enhancing defense against Advanced Persistent Threats, and improving network security responsiveness.
System And Mechanism To Optimize N40 Signaling Reduce Cdr Store In Billing Domain, Mukesh Yadav, Arvind Chandrakar
System And Mechanism To Optimize N40 Signaling Reduce Cdr Store In Billing Domain, Mukesh Yadav, Arvind Chandrakar
Defensive Publications Series
A Charging Function (CHF) enables a control option on the Session Management Function (SMF) where it can consolidate a Used Unit Container (UUC) if multiple UUCs with the same Resource Group (RG) and Service ID (ServID) are present in the ChargingDataReq message. Once this control is implemented, if the SMF sends a ChargingDataReq message with multiple UUCs having the same RG/ServID and SessionCtx, the SMF will merge the usage of these UUCs into a single UUC. This will help reduce the payload size of the signaling from SMF to CHF to Charging Gateway Function (CGF) to the Billing Domain. Additionally, …