Digital Commons Network^™

Refined Music Clustering, Christian Weitenberner, Ullas Gargi, Girum Ibssa, Brian Karlak, Krishmin Rai

Defensive Publications Series

Many different versions of the same song are often found in online music or video services. To enable easier search and better user experience, online music services often use clustering techniques to cluster different versions of the same song. Clustering systems are designed to be tolerant to shifts, noise, small tempo changes, etc. This tends to cause remixes and covers of a song to be clustered with the original versions of that song. This leads to certain problems, e.g., recommender systems selecting the most popular version of a song, in preference to other versions of the song; poor precision-recall rates …

Shush Gesture For Consumer Devices, Scott Jenson

Defensive Publications Series

When individuals within a social group fail to silence their devices during events e.g., meetings, theatrical performances, conversations, etc., notifications from a device can lead to a disturbance to the social group. Such disturbance can result in some pique even in everyday social settings, e.g., a family dinner or a face-to-face conversation.

This disclosure describes use of physical gestures that a user can perform to shush a phone while simultaneously telegraphing an intention to focus to others. The gestures of the user, and the mode of the consumer device when shushed, are such that other members of a social group …

Personalized Ordering For Presentation Of Help Topics, Victor Carbune

Defensive Publications Series

Users often consult online help when they encounter problems. Many online help resources are organized as a list of question-answer pairs or topics presented in a fixed order. Upon user consent, the techniques of this disclosure utilize a machine learning model to personalize the order in which the list of question-answer pairs or topics is presented to the user. The personalized ordering utilizes the knowledge of the context of the user to sort the presented list of help content based on the match of the content with the problem the user is likely facing.

Defending Against Attacks On Biometrics-Based Authentication, Tanmay Wadhwa, Neil Dhillon

Defensive Publications Series

Many devices include biometrics-based user authentication in addition to secret-based authentication. While secret-based authentication involves a precise match with the known secret, biometrics-based authentication involves fuzzy matching that verifies that the input is similar to known biometrics within an acceptable threshold level of difference. As a result, biometrics-based authentication techniques are susceptible to attacks in which a malicious actor attempts to authenticate as the user via biometrics data that is crafted carefully to be similar to the stored biometrics within the threshold.

The techniques of this disclosure guard against such attacks by use of a generative adversarial network (GAN) where …

Matching Language And Accent In Virtual Assistant Responses, Jatin Matani, Philippe Gervais, Marcos Calvo, Sandro Feuz, Thomas Deselaers

Defensive Publications Series

Traditionally a voice-controlled virtual assistant interacts with a user in a neutral, accent-free way. A virtual assistant that matches the language and accent of the user can be more appealing. This disclosure uses machine-learning models to match the language and accent of a virtual assistant to that of the user who commands it.

Automatic Rebooking Of Hotels Using Virtual Assistants, Saptarshi Bhattacharya

Defensive Publications Series

Hotel prices fluctuate based on demand and supply. Hotel bookings are often refundable if cancelled. A user with a refundable hotel reservation can therefore cancel and rebook the hotel if the price drops. Currently, such cancellation and rebooking requires a user to keep track of hotel prices and rebook upon detecting a drop in the price, e.g., from that paid by the user. This disclosure describes an automated method to monitor hotel prices after a refundable booking is made and automatically take action upon detection of a price drop.

Internet Of Things - Data Security And User Authentication Management, Ronald Heiby

Defensive Publications Series

The present disclosure describes systems and methods that provide for managing user authentication and data security for “Internet of Things” (IoT) devices. More particularly, the present disclosure provides for providing an IoT device with a public key portion of a PKI key pair belonging to an authorized user (e.g., owner, etc.) of the IoT device. The PKI key pair of the authorized user can be used to provide user authentication and data security when connecting to and/or communicating with the IoT device. The systems and methods of the present disclosure can provide for improved user control over IoT devices and …

Ground Robot For Charging Multiple Electric Vehicles, Chuck Byers, Ashok Moghe, John Apostolopoulos

Defensive Publications Series

A low profile robotic system moves a Wireless Power Transfer (WPT) primary coil under a row of vehicles in a parking lot. The system positions itself underneath an Electric Vehicle (EV), adjusts the position of its primary coil to align with the receiver coil under the vehicle, and transfers a sufficient quantity of energy to charge the vehicle. Once charging is complete, the robot moves down the row of parked vehicles to charge additional vehicles.

Gantry Robot For Charging Multiple Electric Vehicles, Chuck Byers, Ashok Moghe, John Apostolopoulos

Defensive Publications Series

An overhead gantry robotic system moves a Wireless Power Transfer (WPT) coil across a row of parking spaces in a parking structure. The system detects a vehicle to be charged, adjusts the position of the primary coil over the vehicle, and lowers the primary coil into proximity of a receiver coil in the top of the vehicle.

Active Chassis With Rotary Spring, Daniel Hoppe

Defensive Publications Series

ACTIVE CHASSIS WITH ROTARY SPRING

Customer, Safety And Environmental, Daniel Hoppe

Defensive Publications Series

CUSTOMER, SAFETY AND ENVIRONMENTAL

FUNCTIONS FOR PRIVACY ON DEMAND IN VEHICLE

GLAZING

Use Of A Pyrotechnic Separator For Voltage, Daniel Hoppe

Defensive Publications Series

USE OF A PYROTECHNIC SEPARATOR FOR VOLTAGE

REDUCTION IN THE ISOLATED CIRCUIT

Underbody Protection With Damping, Daniel Hoppe

Defensive Publications Series

UNDERBODY PROTECTION WITH DAMPING

EFFECT

Soc Against Electric Vehicles, Daniel Hoppe

Defensive Publications Series

SOC AGAINST ELECTRIC VEHICLES

Pressure Signal Improving Component, Daniel Hoppe

Defensive Publications Series

PRESSURE SIGNAL IMPROVING COMPONENT

Illuminated Moldings And Frames Radiator, Daniel Hoppe

Defensive Publications Series

ILLUMINATED MOLDINGS AND FRAMES RADIATOR

GRILLE FOR MARKING AUTONOMOUS VEHICLES

Attachment Of A Flag Pole To The Vehicle, Daniel Hoppe

Defensive Publications Series

ATTACHMENT OF A FLAG POLE TO THE VEHICLE

WITH A CONNECTION TO THE FENDER BENCH INSTEAD

OF THE FENDER PANEL (OUTSIDE)

Adaptive Positioning Of Collaboration Device Using Pattern Recognition Of Accessibility User, Damien Mccoy

Defensive Publications Series

A collaboration device may include a large interactive display in the form factor of, for example, an interactive whiteboard. An interactive whiteboard may be a standalone touchscreen computer used independently to perform tasks and operations, or a connectable apparatus used as a touchpad to control computers from a projector. Conventional interactive whiteboards may present challenges for users with mobility or accessibility needs, especially for individuals who use wheelchairs, as they may be unable to reach the interactive whiteboard. Presented herein are techniques that enable an interactive whiteboard to recognize mobility or accessibility impaired users and to dynamically adapt the position …

Triggering An Update Of A Client Application By Selecting A Deeplink, Justin Lewis, Ruxandra Davies

Defensive Publications Series

In some instances, a user of a mobile device may trigger an update of a client application by selecting, from an application other than the client application, a deeplink (e.g., a hyperlink or a uniform resource locator (URL)) that has context and that also links the mobile device to a host program. If the client application is out of date and not compatible with the host program or content provided by the host program, the mobile device triggers an update to the client application. During the updating of the client application, the mobile device maintains the context so that the …

Applications On Top Of Dna Center: Software Facilitated Method To Reliably Isolate, Remove And Insert Devices In The Network, Vishal Murgai, Zach Cherian, Amarender Musku, Ankur Bhargava, Saurabh Agarwal, Nikul Bhatt

Defensive Publications Series

A set of software functions is presented to allow network operators to reliably manage network changes with minimal or no impact to other parts of a currently operating network. Software functions in a centralized network controller such as Digital Network Architecture Center (DNAC) provide network assessment, automation, and mechanism for Graceful Insertion Removal (GIR) feature to communicate with controller for operational reliability. These software functions provide a secured communication channel to an extended version of GIR for easier use along with predictable and reliable network change management.

The present techniques provide a centralized mechanism for a reliable and assured network …

A Logical Layer To Interpret User Interactions, Alison Lentz, Benny Schlesinger, John Thomas Dimartile Iii, Gabriel Taubman, Regina O'Dell

Defensive Publications Series

Some computing user interfaces include a large number of user-selectable options which can make it difficult for a user to locate the right option. Another frequent user interface problem is inadvertent or unintended invocation of actions, e.g., by selection of an incorrect icon from two icons placed close to each other. In some user interfaces, simple actions sometimes require a disproportionately tedious sequence of interactions. This disclosure introduces a layer between the user and the application such that user commands or interactions are interpreted in light of past interactions, and corrected, filtered or automated as appropriate. Past interactions are utilized, …

Blockchain Based Wi-Fi Onboarding Simplification, Identity Management And Device Profiling For Iot Devices In Enterprise Networks, Byju Pularikkal, Santosh Patil, Swami Anantha, Sourav Chakraborty

Defensive Publications Series

A consortium blockchain fabric is provided as a Digital Network Architecture Center (DNAC) Application (APP) stack running in a Cloud Neutral Facility. This consortium blockchain fabric achieves two important functions: Verification of Internet of Things (IoT) Vendor Installed Identity and IoT Device Network Usage Profile.

Verification of IoT Vendor Installed Identity. IoT device vendors will be part of the IoT consortium that maintains the blockchain fabric. The manufacturer installed identity credentials of the devices are registered on the blockchain. This allows the enterprises to verify the manufacturer installed identity of the device at the time of onboarding.

IoT Device …

Weight Compression For Deep Networks Using Kronecker Products, Yair Movshovitz-Attias, Elad Eban

Defensive Publications Series

Deep networks have shown success in many challenging applications, e.g., image understanding, natural language processing, etc. The success of deep networks is traced to the large numbers of neurons deployed, each with weighted interconnections to other neurons. The large numbers of weights result in classification accuracy, but also use significant memory.

This disclosure describes techniques to reduce the number of weights used in deep networks by representing the matrices of deep network weights as the Kronecker product of two or more smaller matrices. The reduction in weights is made possible by the observation that deep networks do not always use …

Active Alignment Method To Dual Balance Camera Spatial Frequency Response, Jamyuen Ko, Chung Chun Wan

Defensive Publications Series

In camera manufacturing, an active alignment process is used to align the optical axis of the lens with the sensor axis of the camera. The active alignment process reduces the spatial frequency response (SFR) corner-to-corner imbalance that causes SFR differences at image corners. The active alignment process is typically carried out with the object at the infinity position with respect to the lens. As a result, in auto-focus cameras with movable lenses, higher SFR corner-to-corner imbalance persists in macro positions. The techniques of this disclosure involve applying the active alignment process at infinity and at macro positions to distribute SFR …

Applications On Top Of Dna Center: Automatic Configuration Validation Based On Topology For Basic Interoperation, Smruti Lele, Ankur Bhargava, Ajay Madhavan

Defensive Publications Series

A rule based service is provided for Configuration Simulation and Validation (CSV) for the devices managed on given network. With an understanding of the architecture of a network, the service can be employed to determine which devices need to be in the same Internet Protocol (IP) subnet. The service can be employed to suggest one or more configuration changes to the administrator such as, changes to address as a potential misconfiguration, improve operation, etc.

Application Of Dna-C: A Method For Short Term Load Prediction And Load Balancing Using Techniques From Chaos Theory, Jerome Henry, Rajesh Pazhyannur, Robert Barton

Defensive Publications Series

Novel techniques are provided for an application built on top of DNAC to predict network load on an access point (AP). These techniques determine a time interval after which each predicted dimension is not accurate, which is reflective of the real load. This allows the system to predict load accurately, without the burden of having to observe all neighboring access points, a computationally expensive process, while being able to determine when new weights and new predictions need to be generated. The result is a computationally-light yet accurate load prediction that can be applied to key Aps (e.g., lobby APs, or …

Applications On Top Of Dna Center: Topology Evolution Based On Customer Network Dynamics, Smruti Lele, Ankur Bhargava, Ajay Madhavan

Defensive Publications Series

Techniques are provided for detecting and managing dynamically changing loads in a network, for example, during large scale events that may cause a bottleneck such as a stadium game event or a musical concert. These techniques allow network administrators to improve network efficiency by monitoring and tweaking the network capabilities in order to handle network load during periods of high demand without a detailed background knowledge of the network.

Lambdaloss: Metric-Driven Loss For Learning-To Rank, Cheng Li, Michael Bendersky, Marc Najork, Xuanhui Wang, Nadav Golbandi

Defensive Publications Series

How to directly optimize ranking metrics such as Normalized Dis­counted Cumulative Gain (NDCG) is an interesting but challenging problem, because ranking metrics are either flat or discontinuous ev­erywhere. Among existing approaches, LambdaRank is a novel algo­rithm that incorporates metrics into its learning procedure. Though empirically effective, it still lacks theoretical justification. For ex­ample, what is the underlying loss that LambdaRank optimizes for? Due to this, it is unclear whether LambdaRank will always converge. In this paper, we present a well-defined loss for LambdaRank in a probabilistic framework and show that LambdaRank is a special configuration in our framework. This framework, …

Providing Link Suggestions Based On Recent User Activity, Jakob Foerster, Matthew Sharifi

Defensive Publications Series

Web browsers typically include a navigation bar to enable users to enter URLs or search phrases. The browser often supplies suggestions to complete a URL or search-phrase when the user types a first few characters into the navigation bar. This disclosure provides techniques to provide link or URL suggestions based on text that was recently copied or highlighted by a consenting user. The text is classified into one or more types, e.g., geographical addresses, product categories, foreign language, etc., using a machine learning model. A map is established between text types and URLs associated with such types. When the user …

A Budget-Based Policy For Operating System Scheduling Of Background Tasks, Alhaad Gokhale

Defensive Publications Series

Background applications in an operating system (OS) can be either running or sleeping. If a background application continues running without going to sleep or attempts to wake up with an unnecessarily high frequency, it needlessly consumes system resources, such as power, memory, etc., thus decreasing operational efficiency.

The techniques of this disclosure allocate each background application with a time budget proportional to the number of CPU cycles. The budget decreases with the use of CPU cycles by the background application until the application goes to sleep or runs out of budget. The budget for each background application is replenished over …