Digital Commons Network^™

Threshold-Free Selection Of Taxonomic Multilabels, Hsin-Yi Lin, Brian Milch, Michel Adar, Scot Fang, Rene Van De Veerdonk

Defensive Publications Series

In online search or content selection systems, significant computational resources are expended to classify or categorize electronic documents into topics, concepts, or entities. A classifier can process, parse or otherwise analyze the document to assign one or more labels to the document based on the taxonomy. The classifier can generate a score for each of the labels, and provide the labels and the scores to other components or modules for further downstream processing. To keep downstream processes efficient without causing excessive processing of labels, the classifier may filter out the labels to return a subset of labels based on comparing …

A Novel Method For Tracking File Based Migration And Its Progress, Faramarz Rabii, Joel E. Lilienkamp

Defensive Publications Series

An efficient and 100% accurate method for tracking when file based migration has completed is presented. The method is based on a simple counting mechanism of FileSystem objects as they are found, and does not require a walk of the remote FileSystem tree prior to the migration.

Client-Side Link Selection Validation, Michael Mcnally, Tuna Toksoz, Alex Jacobson

Defensive Publications Series

A method for reducing unintentional user clicks due to reflow and clickjacking in mobile apps. Links in electronic documents such as web pages may be marked as special or protected. When a protected link is present, the document presentation environment prevents selection of links that exhibit untrustworthy behavior, which might have been clicked by accident in an attempt to select the protected link. Examples of untrustworthy behavior include repositioning the link just prior to the link selection or presenting the link in a transparent manner.

Bloom Filter Locking, Adam Blinkinsop

Defensive Publications Series

Database implementations typically use either fine-grained locking or coarse-grained locking, but not both in the same system. This can lead to greater contention and less efficient operation, e.g., when an application generates queries that have a wide variation. This document describes Bloom filter based locking techniques that are a compromise between coarse-grained and fine-grained locks. The techniques can efficiently handle queries from applications that generate queries that have a wide variation.

Mechanism For Reducing The Cost Paid By End-Users For The Data Consumed To View Ads, Naveen Kalla, Sooraj Sasindran

Defensive Publications Series

A technique is provided for separating ad streams from other services provided to a mobile device, e.g., using a separate socket for the ad stream. This can enable the ability to charge the end-user differently for the ad stream data consumed on the mobile device when the device is on a mobile wireless network that charges for the data consumed, and use lesser bandwidth to stream the ads than that used to provide the other services.

System And Method For Content Ingestion And Optimized Content Delivery, Manikandan Gopalakrishnan

Defensive Publications Series

A system is described for packaging multiple content types into a single package. For example, the package can contain combinations of video content, image content, audio content, and metadata that defines how the types of content are related and used. The package can then be processed and stored, e.g., in a proprietary format with a metadata file that describes how the content is related (e.g., how audio content is to be combined with text content).

Automated Application Permissions Setting, Victor Carbune, Daniel Keysers, Thomas Deselaers

Defensive Publications Series

Operating systems of computing devices include permission management features to grant software applications (apps) access to various hardware and software components. Permissions may be configured using permission sets that each specify different levels of access. A user can specify the level of access to an app by selecting a permission set. A conventional permission set either grants or restricts access to a component. Techniques are described that provide selective access to a component by automatically inferring fine-grained permissions from various user-specific and other available signals.

Generic Mechanism For Transforming A Single-Producer / Single-Consumer Unbounded Linked List Into A Bounded Linked List Without Allocation, Frank Afriat

Defensive Publications Series

SPSC unbounded linked list are very easy to construct and may have interesting properties like Wait-freedom or Lock Freedom but generally provided implementation are unbounded and require Memory Allocation for adding nodes. The purpose of this document is to present an idea giving the possibility to keep all the interesting properties of the Linked List but to transform it into and Bounded Linked List without memory allocation.

Using Content Provider Signals To Select Individual Content Items For A Content Item Campaign, Poonam Suryanarayan, Matthew Burriesci

Defensive Publications Series

This document describes a system and method in which signals from a third-party content provider are used to help select content items in the content item campaign of the third-party content provider. The selected content items are then presented to a resource. The signals may be used to help select between content items for which there are known metrics and content items for which there are no metrics. The signals indicate trends or preferences that may indicate that one content item or type of content item is a better choice compared to other content items.

Automatic Detection Of Video Content, Haskell Garon, Ertürk Diriksoy

Defensive Publications Series

This document describes an assortment of techniques for detecting a media type of uploaded content. Three techniques described above include searching for a content-type tag or a file extension tag in an HTTP header of a get video snippet of code, parsing a DOM tree of a file, and running a JavaScript trace to detect a Mobile Rich Media Ad Interface Definition (“MRAID”) call to the method playVideo().

Optical Circulator For Free Space Optical Communication, Chiachi Wang

Defensive Publications Series

A free space optical communication system transmits and receives optical signals in a colorless manner using an optical circulator. The system installs the optical circulator with a single mode (SM) fiber at port 1, a double clad (DC) fiber at port 2, and a multimode (MM) fiber at port 3. The system injects a first optical signal into a core of the SM fiber. The system then routes the first optical signal at port 1, using the optical circulator, into a SM core of the DC fiber via Port 2. Further, the system injects a second optical signal into a …

Automatic Audio-On Media Content Streaming, Justin Lewis, Richard Rapp

Defensive Publications Series

Disclosed herein is a method for automatically transitioning a video media content stream to an alternate stream that is audio-only when the streaming device connects to an audio-only playback mechanism. The method manifests as a feature within a video sharing application. The application is provided with a setting that automatically transitions the video content stream to an audio representation of the same stream when the client device connects via Bluetooth, Wi-Fi or headphone jack to a speaker device. Similarly, when the audio-only speaker device is disconnected from the client device the application resumes streaming the video stream. The method automatically …

Social Network For Military Veterans, Bernie Hudson

Defensive Publications Series

Many military veterans and service members do not like to use currently available social media channels to stay in touch with other vets they have served with. A social network exclusively for military veterans could help them stay in touch with those they have served with to discuss and help each other on various issues. In this social network, they could form groups, chat, send mail messages, highlight places where they have been deployed to, post pictures and videos of their experiences, and even rate their experiences. They could use all the features available in a typical social media group …

Language Importance Score, Jonasz Pamula, Christoph Oehler, Aleksandar Stupar, Daniel Henell

Defensive Publications Series

A system and method are disclosed for classifying videos based on a Language Importance Score (LIS) indicating how important it is to understand a certain language to be able to understand a given video. The method extracts video features directly from the video after it has been uploaded rather than by analysing viewing patterns from logs. The level of LIS is classified as LOW, MID, or HIGH based on the degree of ease for a user to understand. Global/multilingual video platforms can use this system to optimize search to improve viewer experience effectively.

System For Detecting End Of Speech Utterance For Safe Interruption, Dimitri Kanevsky

Defensive Publications Series

A system and method for detecting end of speech utterance for safe interruption using audio and video processing software or application is disclosed. The system includes hardware that determines the speaker’s intonation in real time or with minimal delay and displays this visually on a screen along with a transcript of the speech delivered by the speaker. The speaker’s intonation and end of speech could be identified in the system display in real-time as rising and falling of pitch at the end of speech. The system and method could be implemented as a module of any video call or conference …

System And Method For Providing Assistive Video/Audio Conference Channel, Liat Kaver Oreamuno

Defensive Publications Series

A system and method are disclosed for providing an assistive video/audio conference channel to a user with hearing assistive devices. The system could be configured to provide a separate channel in a videoconferencing application setting accessible to hearing assistive devices. The channel could be capable of sending audio from any remote speaker. The system and method could be implemented as a virtual additional assistive channel to the existing videoconferencing software applications.

A Mechanism For Making Creation Of Landscape-Oriented Videos More Natural, Scott Atwood

Defensive Publications Series

A method to enable users to naturally create landscape oriented videos by combining the accelerometer function of the mobile device with a slightly modified camera is disclosed. The rectangular sensor of the camera is replaced by a sensor that is square with truncated corners arranged as two overlapping rectangles. While recording the video, the accelerometer will be used to determine the orientation of the mobile device and select which sensor area could be used to record the video. The disclosed device and method allows users to hold their cell phone in the orientation that is most normal and natural for …

Smartphone Personal Assistant For Radio Network Selection During Roaming, Jean-Philippe Paul Cormier

Defensive Publications Series

A smartphone personal assistant system for selecting a radio network during roaming is disclosed. The system makes use of the on-device assistant to understand where the device is located after a “Radio On” or landing from a flight. This information is used in conjunction with a database of frequencies and locations to instruct the radio to scan the correct frequencies. Thus, the system allows the user to operate the mobile device outside the home area network.

Using Boosted Haar Cascades To Detect Rectangular Picture Frame-Like Regions In Images, Sharadh Ramaswamy

Defensive Publications Series

A machine learning system and method with boosted Haar cascades for automatically detecting a rectangular region to replace it by an ad or contents from another image are disclosed. The required features and the logic for detecting rectangular regions are learnt by the system in a suitable machine learning implementation. The method follows the steps of running the pattern recognition algorithm using a Haar transformation on the pixels of an image. The algorithm identifies a first pair of parallel lines separated by a distance, and looks for a second pair of lines in a perpendicular direction to define a rectangular …

Collective Consumer Incentive System, Kamran Shafiei

Defensive Publications Series

A system and process for creating a network of businesses to collectively grant and manage consumer incentives are disclosed. The system allows collaboration and management of joint promotion campaigns between business entities through the Web and mobile technology platforms. The system manages incentives for businesses, their networks and customers. It allows customers to search, find businesses and network with other customers. The customers can view and use incentives that are available to them. The system combines a social network concept for business with customer deals and incentives. It has the advantage of increased customer outreach through cross traffic and referrals …

Braille Reader-Integrated Keyboard, Alberto Martin, Katie Roberts-Hoffman

Defensive Publications Series

ABSTRACT

Systems, devices and methods of providing braille reader-integrated keyboard for use by both visually impaired and sighted users are disclosed. The keyboard is provided with a braille reader of 10 mm or less width, above the function keys of a normal laptop keyboard. Buttons are provided on the side of a regular QWERTY keyboard layout, that allow navigating the menus. The method envisages swapping the existing off-shelf keyboard of any laptop or other computing device for the braille reader-integrated keyboard. The integrated keyboard has several functions including being detectable by the system, and other accessible-friendly features such as audio …

Video Thumbnail Selection Based On Deep Learning, Weilong Yang, Min-Hsuan Tsai, Tomas Izo

Defensive Publications Series

Video thumbnails are often the first thing a viewer sees when browsing or searching for videos. A frame that is visually representative of the video is typically selected and used as a thumbnail representation of the video. Sometimes, such a thumbnail is not an adequate semantic representation of the video. Further, it is possible that such a thumbnail is not visually pleasing. This disclosure describes deep learning techniques to select video thumbnails that are visually attractive and reflect the content of a video. Thumbnails as described in this disclosure are attractive, improve a likelihood of user selection, and help users …

A System For Creating Network Of Businesses To Grant And Manage Consumer Incentives Collectively, Benjami Alexander Bardin

Defensive Publications Series

This disclosure presents a system and method to select unsupervised data for speech processing using the distribution of supervised data in an embedded space. The data sets are represented by different colors to differentiate between supervised and unsupervised utterances. The system samples a set of utterances from the unsupervised data, such that the distribution of the unsupervised sample matches with the distribution of the supervised utterances. The sampling method converts the data sets into bins in a two-dimensional histogram, which is then normalized using the size of the data set for each bin. The data is then manipulated and selected …

Estimating Building Risk Using Imagery And Machine Learning, Michael Greene

Defensive Publications Series

The present disclosure describes systems and methods that use available imagery of a building or other structure in combination with one or more machine learning techniques or machine-learned models to assist in estimating risk associated with the building. The systems and methods described herein can also be adapted to predict failure risk of various types of infrastructure (e.g., utility infrastructure such as power lines and power line poles) and/or industrial assets. Risk assessment can be useful for various objectives including assessment for insurance purposes, valuation purposes, etc. In one example, one or more machine learning models can be used to …

"Method For Improved Bandwidth Utilization In Data Downloading Systems Using Intelligent Dynamic Connection Limit Stepping", Simon Hatch, Oystein Eftevaag

Defensive Publications Series

The subject technology relates to improving bandwidth utilization of data downloading systems by dynamically setting a connection limit with a host. The connection limit (or number of slots) can be dynamically set by a heuristic algorithm that monitors the state of active connections with the host, and steps up or down the allowed number of connections based on what overall percentage of the active connections are currently in a waiting-for-data state. The dynamic connection limit stepping utilizes the connection state in order to automatically adapt to both the available bandwidth and the size of the resources needed to download.

Assessing Brand Visibility Using Machine Learning, Sujoy Banerjee, Michael Greene

Defensive Publications Series

The present disclosure describes systems and methods that leverage one or more machine learning techniques or machine-learned models to perform a large scale statistical assessment of the visibility of an advertising campaign. More particularly, a processing pipeline is provided that takes input imagery and outputs an indication of a visibility of a brand. The pipeline can include a brand recognition neural network and a brand visibility scoring engine. Keywords associated with the present disclosure include: machine learning; neural network; deep learning; model; training; image; imagery; advertisement; advertising; ads; campaign, brand; logo; visibility; effectiveness; score; location; traffic; emblem; trademark; label.

Customer Transaction Voice Recognition Intelligence, Justin Von Kennel

Defensive Publications Series

A system and method are disclosed for more efficient transactions on customer voice calls. The disclosed system and method integrate voice recognition technology into industry standard call center flow. The system monitors and transcribes the call into the customer relationship management (CRM) system and searches the knowledge base using the keywords. The search quickly retrieves relevant data to the agent who provides resolution for the customer’s problem. Voice recognition then completes the CRM request using the transcribed interaction between the customer and the agent. The system and method disclosed will improve customer experience by reducing the time taken for resolution. …

Prediction Of Traffic For Internet Domains, Joya Zuber, Hugues Vincent, Isaac Andres

Defensive Publications Series

This disclosure describes techniques to generate predictions regarding genuine traffic that a domain is likely to receive without access to historical data about the domain. Domain traffic data for known domains is utilized to learn characteristics of domains that are likely predictors of traffic that the domains receive. Characteristics of a query domain are used predict traffic that the query domain is likely to receive. Traffic predictions can be used to identify domains that buy referral traffic and to evaluate domains.

Fast Time Of Flight Camera Lens For Mobile Phone Applications, Anurag Gupta, Ashish Shah

Defensive Publications Series

A lens design for a TOF camera system is disclosed for use in mobile phone applications. The design is physical hardware comprising a four-element camera lens with an image sensor and infrared (IR) filter. The camera lens system, even with a small 9mm diameter, has significantly high performance in terms of contrast, resolution, field of view and working distance.

A Method And Apparatus For Faster, Reliable And Consolidated Logging Of Information During System Failure, Clarete Riana Crasta, Sandeep S Yelandur, Suhas Shivanna

Defensive Publications Series

For servers today, that run mission critical workloads, downtime is not an option and any outage of these servers usually translates to reduced revenue, reduced profitability and potential customer loss. Any interruption in the operation or availability of these workloads will have a ripple effect throughout the organization. Gathering valid and necessary data about the event of failure from all possible sources plays a significant role in determining how quickly and accurately the root cause for the server down-time is identified. The data required for such analysis is spread across Firmware and Operating System (OS) and comes from different sources …