Digital Commons Network^™

A System For Multi-Level, Multi-Variate Embedding Framework For Risk Scoring, Hao Xu, Anurag Tangri, Chiranjeet Chetia

Defensive Publications Series

The present disclosure provides a novel system for multi-level multi-variate embedding framework for risk scoring. In the present disclosure, risk factors within each category are aggregated as one score using Level-1 embedding, while a global risk score is obtained using Level-2 embedding. The framework models different scores considering the interactive effects among different factors and scores. Also, any auxiliary variables that are important for global risk scoring can be easily integrated. The coefficients in multi-level embedding can be determined within one process based on historical data of user interested activities. A penalized loss is used to optimize the model structure …

Automatically Highlighting Browser Tabs Related To Current Tab, Karthik Srinivas, Ajay Prasad, Ramprasad Sedouram

Defensive Publications Series

Users that have a large number of open browser tabs can struggle to identify tabs that are relevant to their current task. While tab search functionality is available in some browsers, such functionality does not automatically identify relevant tabs. This disclosure describes a web browser that, with user permission, determines the user’s current context based on the currently active browser tab and automatically highlights other relevant tabs that are currently open. The content and metadata of browser tabs are analyzed, e.g., using on-device machine learning techniques, to identify the relevant tabs. The relevant tabs can be highlighted with different colors …

Approaches For Incorporating A Variety Of Metadata In Transformer Operation, Antonio Gulli, Afroz Mohiuddin, Samer Hassan

Defensive Publications Series

A plain transformer model typically leverages only one piece of metadata - position encoding - directly in the transformer model. The use of transformers typically involves expensive and complex external scaffolding before or after output generation to avoid issues such as hallucination, irrelevance, etc. This disclosure describes techniques to incorporate a variety of metadata types into the native architecture of transformer models. The additional signals can help avoid hallucinations, improve relevance, and minimize the need of expensive external scaffolding. Generalizing transformer operation to incorporate a diversity of metadata can be achieved in various ways such as adding a metadata embedding …

Identifying Aggregators Systematically Using Transactional Attributes (Card Acceptor Id Or Caid), Basudeb Ghosh Visa, Prithwiraj Mitra Visa

Defensive Publications Series

A method, system, and computer program product are provided for identifying aggregators using transaction data. The system may include at least one processor programmed or configured to receive transaction data comprising a card acceptor identifier (CAID) and a card acceptor name. The at least one processor may identify a plurality of CAIDs included in the transaction data and for each CAID, determine a number of distinct card acceptor names corresponding to the CAID. The at least one processor may then identify a first CAID having a plurality of distinct card acceptor names corresponding to the first CAID. The at least …

Sensory, Faiz Ahmed Visa, Sudarshan Krishnakumar Gaddam Visa

Defensive Publications Series

The methods and system disclosed in present disclosure provides a generic sensor library which comes with some generic sensors and also provides a default controller to combine the sensor and the respective action that need to be performed. The sensor library provides basic checkpointing mechanism to provide the restart capability to the sensor. Though the sensor library provides the defaults which is applicable for most of the cases but user has the option to extend as well as implement their own sensors, checkpointing mechanism, trigger, etc. and use the default controller with these custom components. The present disclosure provides the …

Voice-Activated Selfie

Defensive Publications Series

The subject disclosure provides for an extended reality (XR) wearable device that may capture a mixed reality self-portrait and share the self-portrait to a social media platform. The disclosure concerns the problem of conveniently capturing and sharing digital self-portraits. The disclosed solution addresses the problem by configuring a set of XR smart glasses to, based on a voice command of a user, capture raw real, virtual, or mixed images; process the raw images into a self-portrait that includes real, virtual, or mixed elements; and share the self-portrait to a social media platform.

Method And System For Secure Credential Generation, Rhidian John

Defensive Publications Series

One embodiment includes a method performed by an authentication provider. The method includes receiving a sensitive information from a user. For example, the sensitive information may include payment data or a healthcare data of the user. The method includes transmitting the sensitive information to a processing server and receiving a long living token associated with the sensitive information from the processing server. The authentication provider stores the long living token. The method further includes the authentication provider receiving, from a resource provider, a request for a subset of the sensitive information. The method includes transmitting the subset of the sensitive …

Generative Ai Driven Content Quality Enhancement For User-Generated Content, Jaunani Sriramachandran, Ramprasad Sedouram

Defensive Publications Series

Online content posted by amateur content creators may not be of sufficient quality to drive engagement. For example, such content may have audio or video glitches. Viewer feedback can indicate the problems with user-generated content. This disclosure describes techniques that analyze viewer comments and/or the user-generated content to generate prompts for a generative AI model. With permission from content creators, the capability of generative artificial intelligence models to execute audio-enhancement tasks such as volume equalization, noise reduction, audio filtering, etc. and video-related tasks such as video smoothing, video stabilization, etc.is utilized via the generated prompts to automatically improve their content. …

Improving Online Presentations Based On Audience Usage Of Conversational Assistant, Dharma Teja, Ramprasad Sedouram

Defensive Publications Series

During a presentation made via video conference, attendees may use a conversational assistant within the video conference to seek clarifications for terms or concepts that they don’t understand. Speakers that are unaware of such audience activity cannot adjust their presentation to clarify presentation content. This disclosure describes techniques that enable a presenter to receive live feedback regarding content of their presentation that may not be understood (or may be misunderstood) by the audience and provide clarifications in real time. During the presentation, a presenter's interface is augmented with indications that a portion of the audience is having difficulty understanding certain …

Content-Aware Power Management For Emissive And Non-Emissive Displays, Daniel Solomon, Manali Bhutiyani

Defensive Publications Series

High display power consumption can lead to reduced battery life and/or device overheating. While reducing display brightness can mitigate these issues, lower brightness can make the displayed content harder for a user to view and reduces the quality of user experience. This disclosure describes techniques that enable specifying precise power budgets for an electronic device with emissive or non-emissive display. The precise power budgeting optimizes power consumption while maintaining display and visibility. Per the techniques, the power consumption of a display is determined in real time and the display brightness is dynamically adjusted to keep display power usage within a …

Content-Aware Sharing Capabilities, Soya Li, Jieyu Tan

Defensive Publications Series

During a sharing session within an online meeting application, a piece of content may be presented, delivered, etc. by a sharing user and viewed, listened to, etc. by receiving users. Techniques are presented herein that support the application of artificial intelligence (AI)-based analysis to such content to develop content-aware sharing capabilities that enable a receiving user to more accurately offer feedback (comprising, for example, a verbal discussion, a submitted reaction, a text-based chat, etc.) with the display of the same occurring in a more immersive and precise fashion.

Providing An Eco Mode Feature For Optical Modules, Antonio Carugati

Defensive Publications Series

Techniques are presented herein that leverage the reality that many optical link installations could safely operate even with a reduced transmission optical power level. Aspects of the presented techniques offer a new optical transceiver feature (which may be referred to herein as an 'Eco Mode') in which a module may be set to work with a lower transmission power level. Under such an approach, at least one extra power level is allowed beyond (i.e., in addition to) an existing nominal power level. Through an Eco Mode option, a 100 gigabits per second (G) transceiver that is deployed in a Quad …

Aligning Map Patches To Estimate Lane Marker Geometry, Daniel J. Lingenfelter, Stephen P. Russell, Simon Desai, Jonathan T. Barron, Daniel Sievers

Defensive Publications Series

Accurate lane marker geometry is important for autonomous vehicles to navigate roads safely and efficiently. Map patches provide valuable data for estimating lane marker geometry but need to be aligned before they can be used to estimate the lane marker geometry. Traditional feature-based approaches for aligning map patches are unsuitable for map patches that include only lane markers, which typically lack distinctive features. This disclosure describes techniques for aligning map patches, specifically aligning lane markers on road geometries. Per the technique, correspondence between lane markers is determined, focusing on smooth curves without distinct features. Map patch alignment technique combines sensor …

Robust Anti-Deepfake Measures For Audio Using Blockchain And Machine Learning, Stephane Belmon, Roshan Rammohan

Defensive Publications Series

Deepfake audio is fake audio where the content, while sounding like it is in the voice of a particular speaker, is not actually spoken by the person. With the advent of deepfakes, verifying audio as being provided by a particular person, e.g., a public figure or other party of interest, and at a particular time and location is important to promote trust within a society. This disclosure describes techniques that enable the listener of a particular piece of audio to verify that the audio was indeed created by the person associated with the voice. A recorder that is certified as …

Occupancy-Based Environment Control For Physical Spaces, Jaime Guajardo, Dmitrii Boiarshinov, Gabriella Lanning

Defensive Publications Series

Optimizing environmental factors such as air ventilation flow, temperature, lighting levels, etc. with time-based fixed settings or presence detection by motion sensors can waste energy and provide less than optimal conditions for users. This disclosure describes techniques that use information obtained (with user permission) from multiple available sensors to optimize the settings for various environmental factors such as light intensity, temperature, air ventilation, etc. within public or private spaces. Sensor observations from smart home devices/ smart appliances within the physical space, as well as from mobile/ wearable devices such as smartwatches, smartphones, etc. are combined to detect human presence within …

The Method To Let Consumers Select The Required Items With A Sense Of Realism And A Sense Of Trustworthiness Utilizing Virtual Viewing Technologies, Punarjeewa Abeysekera

Defensive Publications Series

This paper describes a method to let consumers select the required items with a sense of realism and a sense of trustworthiness utilizing virtual viewing technologies. The method incorporates a website which serves as an index for the available items, a collection of websites belonging to the shops where the items are physically stored, a collection of stereo cameras and a headset which is capable of viewing virtual images. Utilizing the method, the consumer will be able to obtain a collection of multiple 3D stereoscopic perspective images of the actual physical item of interest that is physically stored in the …

Producing And Applying Runtime Adjustments To Semantic Textures, Omar Galal El Sheikh

Defensive Publications Series

Semantic textures are labels associated with objects or regions in a scene, map, or video, and they enable the accurate interpretation of geospatial content. Auto-generated semantic textures frequently include mislabeled regions, e.g., water labeled as ground (and vice-versa). This disclosure describes techniques to enable the touching-up (correcting of semantic labels) of two- or three-dimensional maps, scenes, or videos such that the touch-ups precisely correspond to real-world locations. Touch-up edits can be executed by painting over existing semantic textures using lat-long-centered, touch-up texture tiles as paint brushes within a photo-editing software. The parameters of the touch-up tile, e.g., its resolution, its …

Reconstructing A Machine Learning Model Based On Thresholded Model Differences, Nam Nguyen, Jeffrey Hui, Aroma Mahendru

Defensive Publications Series

A machine learning (ML) model often comprises many floating point numbers (model weights) and the operations applied on them (the computation graph). If a model is derived from another, the derived model often has similar weights and is of the same size. It is a waste of storage to store copies of the similar weights and computation graphs for two related models. This document describes techniques to obtain a sparse representation, referred to herein as thresholded model diff, that can be applied to a base model to reconstruct a version of a derived model. Differences between weights of the base …

A Method And System For Authorizing User Information Using A Passport Wallet Service, Salvador Ricardo, Jenny Ramirez, Cristina Villarroel, Liset Iturriaga Chevez, Alfredo Iglesias, Maikel Zuniga

Defensive Publications Series

The present disclosure relates to a method and system for authorizing user information using a passport wallet service. In an embodiment, the present disclosure suggests registering user information and user passports in a passport wallet service via a merchant device/mobile application. After successful registration, the mobile application transmits a request to a VPVP token provisioning service to generate a secure token associated with the user registration in the passport wallet service. Thereafter, the VPVP processes the request and transmits a unique token to the mobile application. The user reviews the list of the available tokens in the mobile application and …

Spend Smart, Palakh Shangle Visa

Defensive Publications Series

The present disclosure relates a Digital Receipts Platform system to transform the way users manage their transactions. The Digital Receipts Platform system allows users to store, organize, and analyse all their purchase receipts in a unified, digital platform. By integrating with various points of purchase, such as online stores, physical retailers, and app-based services, the platform may automatically retrieve and store digital receipts. The platform provides the users with a comprehensive view of their spending habits and make managing expenses more efficient. It may help users to understand their financial behaviour better and provide useful insights to improve their financial …

Suspicious Call Detection And Mitigation Using Conversational Ai, Kolati Mallikarjuna Rao, Bhavikkumar Patel

Defensive Publications Series

Spam or scam calls and messages are an annoyance, pose security risks, and can harm users that fall prey to calls that request money transfer or other action. Detection of such calls is difficult as scammers evade detection through tactics such as changing numbers they call from, modifying the call script, etc. While some smartphone applications can detect such calls or text based on caller ID, call origination, etc., such techniques cannot easily adapt to new scams. This disclosure describes the use of a conversational AI model to detect suspicious calls. The conversational AI model is trained on a dataset …

Data Lineage Resolver For Grounding Large Language Model Inference, Hari Bhaskar S

Defensive Publications Series

An important challenge when using LLMs in business applications is to account for different sources of information that have conflicting values for the same data at the time the LLM performs the inference task based on such data. When multiple potentially conflicting sources of data are available, LLMs used in business applications may need grounding with facts. This disclosure describes a data lineage resolver that can utilize a data lineage catalog to enable LLM output to be verified for grounding based on the data source(s) used for the output and once verified, provided to the business application. The data lineage …

Container-Based Isolation For Managed On-Device Machine Learning, Hari Bhaskar S

Defensive Publications Series

Bundling ML models can compromise model security and tightly couples model execution with application execution. Managing computing resources can be difficult if multiple applications access the same model and maintaining multiple ML models through their life cycle can become increasingly complex as the application grows. This disclosure describes techniques that utilize a function-based paradigm for on-device machine learning models. The functions can be implemented using various ML models, implemented in isolated containers. Per the techniques, on-device applications (e.g., mobile apps) can be defined with IAM/roles to access one or more functions that provide ML functionality. The techniques provide enhanced security …

Generating Cloud Resource Descriptions Using Generative Artificial Intelligence, Paras Mamgain, Sameer Adhikari, Hariharan Balasubramanian

Defensive Publications Series

The description fields of networking resources within cloud services are optional, and, as such, developers often leave description fields empty. Description fields can be important when the cloud service faces an outage or malfunction. An empty or out-of-date description field is of little use to personnel trying to determine whether the resource under consideration is a cause of the malfunction. This disclosure describes the use of generative artificial intelligence to automatically generate description fields for cloud resources leveraging aspects of the integration of a network resource with other components in a cloud service. Detailed descriptions and descriptive names can be …

Slider User Interface With Dynamic Non-Linear Expansion For Precise Value Selection, Raviteja Govindaraju, Siyan Ks, Ramprasad Sedouram, Ajay Prasad

Defensive Publications Series

Sliders in user interfaces allow value selection from a set range, typically with fixed increments. However, precise selection beyond preset increments is often unachievable due to space and sensitivity constraints, forcing users to use keyboard inputs for exact values. This disclosure describes an enhanced slider that permits fine-tuning of values via touch or gesture input without needing a keyboard. Based on user input such as a press-and-hold gesture, the slider scale is dynamically expanded, and the increment associated with slider travel is adjusted to enable input of precise values. The described slider can be incorporated in any user interface that …

Dynamic Headroom Control For Audio Amplifiers Using Iv Sensing, Chintan Trehan, Todd P. Marco, Dayu Qu, Tao Lin

Defensive Publications Series

In a conventional amplifier and speaker setup for audio playback, a fixed voltage power supply, determined by the expected maximum peaks in input audio signal as well as a buffer value (headroom), is used to avoid distortion. A higher supply voltage is necessary to ensure sufficient headroom when the input audio has high peaks. A fixed headroom can cause high switching losses, reducing overall amplifier efficiency. This disclosure describes techniques to improve the efficiency of class-H amplifiers by dynamically changing the headroom based on distortion observed in audio output. A smart audio amplifier uses IV sense technology to measure the …

Thermal And Performance Efficiency Optimization By Characterization Curve, Hp Inc

Defensive Publications Series

Performance, thermal and acoustic are key features on Gaming PC. The existing solution in ODM/OEM SW UI is to provide the “Performance Mode”, allow user to turn the default (Balanced mode) setting to Performance mode via increasing processing unit (e.g. CPU, GPU) power consumption. The pain point of performance mode is power, temperature & fan noise increasing, but the performance efficiency is Not exactly linear incremental, we need to find out the best & optimized efficiency “operation point” for performance mode.

The new approach is using power & thermal characterization process on user’s PC environment or product factory to determine …

Visa Rtp Cards, Rajkumar Baskaran Visa

Defensive Publications Series

The techniques of the present disclosure utilize the soundbox or any embedded device (e.g., a POS machine) to facilitate RTP in the UPI payments system using contactless cards. In the present disclosure, the consumers are provided with the contactless cards by issuer banks just like normal payments cards (e.g., debit/credit cards). The consumers may apply for the contactless cards via 3^rd party payment applications or directly from banks. In the present disclosure, a contactless card reader and a number pad is attached with the soundbox or the embedded device, where the merchant enters the payment amount for static QR …

The Configuration Which Enables To Experience A Vertically Narrow, Horizontally Wider Virtual Image Projection Area In The Field Of View, Punarjeewa Abeysekera

Defensive Publications Series

This paper describes a configuration which enables to experience a vertically narrow, horizontally wider virtual image projection area in the field of view. The configuration consists of two identical projection units. Each projection unit contains two overhead video displays, a beam splitter, a convex lens part and a concave mirror part. The configuration utilizing the two identical projection units, will have the capability to project virtual imagery onto the field of view of the user, where the virtual image projecting area will be a vertically narrow, horizontally wider space.

Usage-Based App Ordering In Application Switching User Interface, Alex Chen, Ankit Mohan, Yuan Li

Defensive Publications Series

Switching between applications is a frequent activity. Currently, the list of applications available to switch to is sorted based on recency of use, requiring the user to perform complex navigation actions to locate and switch to a target application. Further, the list excludes apps that are not currently running. When a large number of apps are running, such a recency-based display of apps is inconvenient, especially if the user intends to switch to an app that was not used recently. This disclosure describes techniques, implemented with user permission, to personalize the sort order of apps that are displayed when a …