Digital Commons Network^™

Faster Cluster Convergence And Liveliness Detection Through Hierarchal Bidirectional Forwarding Detection, Tapraj Singh, Jeffrey Lo

Defensive Publications Series

Techniques are described herein for hierarchal Bidirectional Forwarding Detection (BFD) to scale liveliness detection. Scalability of BFD sessions to Master and Standby nodes is a concern in current implementations. A fast detection mechanism such as BFD is required to support fast detection of liveliness of cluster nodes.

Automatic Correction Of Disfluent Spoken Queries, Manaal Faruqui, Siddharth Gopal

Defensive Publications Series

A user’s interaction with a virtual assistant typically involves spoken requests, queries, and commands which often includes disfluencies. This disclosure describes techniques to automatically correct disfluent queries. Per techniques of this disclosure, a disfluency correction machine learning model is utilized to convert a disfluent query to a corresponding fluent query. Lexical features extracted from the disfluent query are utilized to determine a portion of the query that is removed from the disfluent query to convert it to a fluent query. The model is trained using pairs of queries.

Distributed Location Processing Using Reverse Tracking And An Internet Of Things Platform, Jeff Apcar, Vinay Saini, Ankush Arora

Defensive Publications Series

The concept of location tracking is of great relevance in various enterprise and industrial environments. The techniques presented herein propose location tracking techniques that are independent from centralized servers and that use scalable technology and architectures to provide resiliency, faster location updates, reduced cost, and increases reliability.

Peer User Approval Based Binary Whitelisting, Josh Zukoff, Russell Hancox, Ed Eigerman, Matt Doyle, Matthew Suozzo, August Huber, Ben Grooters

Defensive Publications Series

Enterprises face challenges in monitoring execution of software binaries. This disclosure describes social voting for enterprise level binary whitelisting. Per techniques of this disclosure, a peer user driven approval process is utilized for binary whitelisting. At a time of launch of a binary that is not pre-approved, a user is provided with information associated with the binary and directed to the social voting process. The user designates a peer user and requests that the peer user approve execution of the binary. The peer user is provided with information about the requesting user and about the binary. Approval by the peer …

Run Time Translation Of User Interface Text Via An Operating System Translation Service, Steffen Meschkat

Defensive Publications Series

User interface text labels are translated into multiple languages to support users who speak the corresponding languages. Software developers use translation services to generate translations of each label in the UI in multiple languages. The translated labels are packaged with the application when it is built. This approach provides a less than satisfactory user experience since the same text label may be translated differently in applications from different vendors. Further, user preferences are not taken into account when generating translated text labels. The techniques described in this disclosure generate translated labels by taking into account user preferences and serve to …

Device-To-Device Charging Through The Display Screen With Seamless User Interface, Vijayasekaran Boovaragavan, Christopher Baker

Defensive Publications Series

Battery life for electronic devices is limited and users do not always have easy access to fixed or mobile power sources to recharge. The location of a power source may also limit the user’s ability to easily use their device. Users may start a task on an application on one device, but then want to switch to another device to continue the task as they change context.

This disclosure describes technology that enables a first device to be placed on the display screen of a second device to charge the first device wirelessly. A wireless charging transmission coil is provided …

Using Computer Vision To Label And Search A Physical Space, Hauke Heibel, Damon Kohler, Stefan Hinterstoisser, Martin Bokeloh, Jürgen Sturm

Defensive Publications Series

Effective operation of a warehouse requires keeping track of the location of various assets within the physical environment. As various sensors are carried through the warehouse environment by operators, range data collected by the sensors over time can be used to reconstruct 2D and 3D representations of the space. This disclosure describes techniques to estimate the locations of Point-Of Interest (POIs) and Regions-Of-Interest (ROIs) within a physical environment such as a warehouse. The location estimates are generated using a combination of 2D visual search of images containing text labels and barcodes, 2D/3D environment reconstruction using sensor data, and estimated trajectory …

Smart Email Send Scheduling, Ashish Duggal

Defensive Publications Series

Although users can select a time and date to schedule an email, such selection is currently done manually. The inconvenience of manual scheduling is accentuated when there are multiple recipients or when a slot for transmission of email is to be selected across time zones. This disclosure describes techniques that automatically schedule the transmission of email based on available information pertaining to the recipients, e.g., information available on recipient calendars, on office productivity software, etc.

Zif Connector With Open End For Fpc Pass-Through, David Rosen

Defensive Publications Series

A zero insertion force (ZIF) connector is a commonly used electrical connector that makes a connection between a flexible printed circuit (FPC), which is a type of flexible cable, and a printed circuit board (PCB). Due to part and assembly tolerances, the FPC is typically oversized (made longer) to ensure its electrical connectivity to the ZIF connector. As a result of the oversizing of the FPC, and due to the ZIF connector not being able to pass through the FPC, the FPC bends or buckles, forming a loop. The loop causes a loss of space within a device that uses …

Supply Chain Trust As A Service Using Trust Management In Blockchain And Internet Of Things To Provide Reputation Scores For Supply Chain Entities, Johnson Manuel-Devadoss

Defensive Publications Series

Techniques are described for Supply Chain Trust as a Service (SCTaaS). SCTaaS may provide trust and reputation scores for supply chain entities using trust management in a blockchain and the Internet of Things (IoT).

Selective Notifications Based On Location And Calendar Information, Jeff Palm

Defensive Publications Series

This disclosure describes techniques to notify individuals that are likely to have been present at a particular location, e.g., a meeting room, based on calendar information, obtained with user permission. For example, a corporate campus can include several meeting rooms with associated calendars that are used for meetings. A person that finds a forgotten item in a meeting room can access a portal to report the item, along with the location and time that the item was found, e.g., via a portal. The portal can then access, with user permission, calendar information of the meeting room and/or other individuals to …

Fast Adaptive Voltage And Boost Frequencies For Central Processing Units, Derek Basehore, Nick Sanders, Stephen Boyd, Sonny Rao, Alex Levin

Defensive Publications Series

This publication describes methods, techniques, and apparatuses that enable a user equipment (UE) to quickly increase or lower the supply voltage and/or the clock frequency to handle changes in load operating conditions of the components of a system on chip (SoC). The UE uses a dynamic voltage and frequency scaling (DVFS) to handle changes in load operating conditions. During the DVFS, an application processor (AP) writes the supply voltage and the clock frequency settings to shared memory between the SoC, the AP, and a microcontroller unit (MCU). The MCU, then, can change the supply voltage using a voltage controller and/or …

Enhanced Telephony Using Human Inaudible Data Over Voice Channel, Asher Segel-Brown, Stefan Zinke-Allmang, Honna Segel

Defensive Publications Series

A telephone, such as a mobile phone, places a telephone call to another telephone by transmitting a signal (e.g., digital or analog) indicative of a sound wave. The signal indicative of the sound wave encodes sound data over a voice channel. The sound data includes human audible data, such as data indicative of sounds (e.g., voice) captured by a microphone. That is, the frequency of the sound wave that encodes the human audible data is within a frequency range audible by humans. The sound data also includes human inaudible data, such as a telephone number of the calling device, an …

Automated Image Retargeting At Scale Using A Generative Adversarial Network, Jingbin Wang, Ramin Zabih, Chen Wang, Richard Strong Bowen, Michael Wilber

Defensive Publications Series

Given the diverse variety of device screens, approaches that rely on manual creation of various forms of an image for each display situation do not scale. Therefore, automated techniques are required to process and resize images to adjust for different devices on which the image is to be displayed. The goal of automated retargeting is to change the aspect ratio of the original image while preserving the semantic and visual meaning of the content within the image. Currently, typical retargeting processes involve a multitude of operations, use a variety of heuristics that are tuned manually, and do not work effectively …

Spoof Detection For Fingerprint Sensors, Firas Sammoura, Jean-Marie Bussat

Defensive Publications Series

This publication describes methods, techniques, and apparatuses that can achieve spoof detection in fingerprint sensors by confirming that a person is present and authorized to gain access to a mobile device, door entrance, vault, application software, physical locations, and/or or virtual activities that the person wants guarded. To do so, machine-learned models may be used to match at least one of a fingerprint residual on the glass covering the fingerprint sensor to a live fingerprint, a fingerprint coloration due to the hemodynamics of the pad of the finger, or a distortion pattern of the fingerprint due to the rotation of …

Virtual Assistant For Planning And Booking Holiday Trips, Miglė Padegimaitė, Sandro Feuz

Defensive Publications Series

Booking and planning holidays is a time consuming activity that is performed multiple times by most users. This disclosure provides a virtual assistant that intelligently simplifies holiday planning and booking. A discovery and booking flow specifically designed for booking holidays is implemented to offer multiple options that the user can choose from at each step, or a limited set of complete plans are displayed based on user requirements determined based on user context, obtained with user permission.

Automatic Design Of Neural Architectures For Recommendation And Ranking Tasks, Anonymous

Defensive Publications Series

Designing the right neural network architecture for a given machine-learning task is critical for performance. For example, the most appropriate neural networks for tasks such as image classification, speech recognition, click-through-rate prediction, etc. are different from each other. This disclosure describes a framework for conducting searches for neural architectures that perform recommendation and ranking tasks.

Counterfeit Good Detection Using Radar, Nikesh Tayi Dhar, Jaime Lien

Defensive Publications Series

Radar, such as frequency-modulated continuous wave (FMCW), can be used to analyze characteristics of an item. The radar can be emitted by a mobile device, such as a smartphone, that has radar technology incorporated as part of it. The radar backscatter of various portions of an item can be analyzed to determine whether the item is likely counterfeit or authentic.

Preprocessing Logic To Enhance Videotelephony Services During Suboptimal Network Conditions, Chinmay Dhodapkar

Defensive Publications Series

This publication describes processes and techniques, implemented on a computing device, directed at enhancing videotelephony services during suboptimal network conditions. In an aspect, a several-stepped process employs a machine-learned technique through which captured frames (e.g., images captured by a videotelephony application) can be segmented and altered, such that an object of interest can be presented with a static background. The entirety of this process referred to herein as preprocessing logic, occurs prior to encoding.

Augmented Reality Immersion Adjustment, Lachlan "Locke" Dunn

Defensive Publications Series

This disclosure describes techniques to adjust the immersion level in an augmented reality (AR) environment. An input mechanism, e.g., a dial in the form of a wheel, a slider, buttons, etc. is provided on an AR device. The user can operate the dial to adjust the level of immersion in the AR environment. The AR environment is rendered based on the selected immersion level. The user can designate services and augmentation desired at each level of immersion, or such determination can be made automatically.

Change Detection For Optimized Semantic Video Analysis, Elly Nattinger, Austin Mccasland

Defensive Publications Series

Semantic analysis or annotation of videos is most useful when done frequently enough to capture the significant moments of a video, but not so frequently that annotations become busy and repetitive. With current techniques, semantic analysis is done too often, overloading the semantic analyzer and overwhelming the viewer with frequent, repetitive, or similar annotations of insubstantially differing frames. This disclosure presents techniques that detect substantial changes in the video for the purposes of semantic analysis. Timely and relevant annotations are presented to viewers without overwhelming them and without overloading the semantic analyzer.

Coupling Thermal Subsystems On Head-Mounted Displays, Matthew Valente, Srivatsan Ravindran

Defensive Publications Series

A multi-body augmented or virtual reality device is a head-mounted display whose constituent units can be flipped up, e.g., from over the eyes to over the forehead. In a multi-body AR/VR device, thermal problems associated with high processing power are exacerbated by the need to cool down each constituent unit. This disclosure describes techniques that achieve efficient thermal regulation in a multi-body wearable device by establishing a conductive or convective thermal path between the constituent units of the device. The techniques enable a flexible and ergonomic industrial design, and reduce duplicated componentry, weight and cost.

Stabilizing Augmented Reality Reticles, Xavier Benavides Palos, Bryan Woods

Defensive Publications Series

Augmented reality applications place virtual objects in a user's real, three-dimensional space. To do so, such applications use reticles, e.g., crosshairs within the screen of the mobile device, to enable the user to select a placement point in real space. The reticles are placed in the image of the real space by ray-casting from the center of the screen to planes detected in the real space. Due to a constant, low-amplitude, involuntary, noise-like movement of the user’s hand, the placement of the virtual object is often imprecise and difficult to localize. This disclosure describes techniques that enable a user to …

Multifactor Technique To Bookmark And Shop Items From A Screenshot, Diane Wang, Aiko Nakano

Defensive Publications Series

Users often take screenshots or photos of items that they like, with the intent to further review and/or purchase those items at a later time. This disclosure utilizes computer vision techniques to parse a screenshot, e.g., to identify image content. Metadata associated with the screenshot is also determined. The information thus obtained is utilized to enable a user to continue shopping at a later time by starting from the screenshot. A deep link to the identified item from a particular vendor, price information, similar styles, other stores that sell the item, etc. are determined and are presented in a user …

Robust Adhd Testing By Applying Clustering Techniques To Survey Responses Or Speech Data, Lingyi Zhang, Cliff Chin Ngai Sze

Defensive Publications Series

Existing tests for attention deficit hyperactivity disorder (ADHD) may exhibit some bias. Also, these tests require filling in a survey with subjective responses, which can lead to misdiagnosis. The techniques described herein reduce bias in ADHD tests by seeking clusters in test-parameter space conditioned on certain characteristics of a person. Clustering is performed using machine learning techniques. With user permission, speech data is obtained via one or more devices such as a phone, smart speaker, etc. an is used to make objective diagnoses of ADHD.

Object Recognition Based Contextual Responses To User Queries, Aiko Nakano

Defensive Publications Series

This disclosure describes techniques to identify specific objects using image analysis techniques and providing contextual responses to user queries based on the recognized objects. The user query is parsed and matched to available information about the recognized object to provide a contextually appropriate response. In addition to object recognition, text recognition can also be performed to obtain information about the object. The described techniques can provide answers to various users that may otherwise be unable to understand how to use an object, e.g., due to inability to read a language, low vision, or other reasons.

Dynamically Generated And Targeted Augmented Reality Shopping Campaigns, Aiko Nakano, Diane Wang, Max Spear

Defensive Publications Series

In augmented reality (AR) shopping, a user points a mobile device camera at a specially-designed poster (or other object) and receives information or shopping deals. Today, AR shopping content, e.g., QR-code-based AR or time-triggered ad filters, is largely templated, static, and one-to-one between a specific target and a specific result. A gap exists between users looking at an ad and actually making a purchase. This disclosure describes techniques that enable a suite of dynamically generated and targeted AR shopping campaigns that directly link to products based on inventory conditions. Facility is provided for contextual deals, e.g., location-specific, time-based, and/or chance-based …

Adjusting The Rotation Speed Of A 3d Object According To Zoom Level, Xavier Benavides Palos, Anshuman Kumar

Defensive Publications Series

When a user attempts to rotate a 3D object in a 3D viewer while the object is scaled up at a high level of zoom, the speed of rotation can be too fast to allow the user to discern and focus on specific parts of the object at that size. This disclosure describes techniques to adjust the speed of rotation of an interactive 3D object depending on the scale at which the user is viewing the object. The dynamic adjustment of the speed of rotation make it possible for users to look at the details of the object at various …

Operational Rules To Manage Power Consumption For Content Display, Katie Leah Roberts-Hoffman, Gregory Granito

Defensive Publications Series

The display is often responsible for a large proportion of the overall power consumed by a device. The total power consumption of a display depends on a number of factors such as the number of pixels that need to be switched on, the color and brightness of the pixels, and the rate at which the on-screen content is updated. Displays that draw high amounts of power can deplete the available power resources of a device relatively quickly, thus shortening the amount of time a user can operate the device without needing to charge or replace the device battery. This disclosure …

Approximating 3d Poly Mesh Virtual Objects With Primitive Shape Colliders, Chloe Snyder, Alper Gungormusler, Andrew Allen, Aleksandr Ayvazov, Dobromir Yordanov

Defensive Publications Series

A seamless immersive experience requires appropriate physical interactions between 3D objects in the scene being viewed. Systems typically handle such interactions by detecting and resolving collisions between 3D objects in the scene. However, collision detection between two high poly meshes of the two colliding 3D objects has the computational order of O(N²) where N is the triangle count for the respective meshes, thus making the computations time-consuming which prohibits real-time use. Current techniques include manual creation of simpler collider shapes such as spheres and boxes to approximate the geometry. This disclosure describes techniques that automatically approximate the poly …