Open Access. Powered by Scholars. Published by Universities.®
- Publication Year
- File Type
Articles 9901 - 9930 of 11801
Full-Text Articles in Entire DC Network
Combining Domain Keywords Into Automatic Document Classification By Deep Neural Network, Qian Diao, Abhijit Talathi, Sarah Kuppert, Robert Frost, Adam Mattern
Combining Domain Keywords Into Automatic Document Classification By Deep Neural Network, Qian Diao, Abhijit Talathi, Sarah Kuppert, Robert Frost, Adam Mattern
Defensive Publications Series
Techniques are described herein for combining the domain keywords into an automatic document classification system. It uses a Deep Neural Network (DNN) model that not only captures the word dependencies and context information in the document, but also encodes the domain knowledge into the category decision process efficiently.
Thermal Downlink Throttle Ue Specific Approach, Blake Kragten, Shivank Nayak, Siddharth Ray
Thermal Downlink Throttle Ue Specific Approach, Blake Kragten, Shivank Nayak, Siddharth Ray
Defensive Publications Series
During operation, temperatures of a mobile communication device of an end user (user equipment or “UE”) can increase to high levels. In particular, UE temperatures can rise when receiving large amounts of data during downlink communication with a network. Generally, there is no ability to limit the amount of downlink throughput when the UE device is in single carrier mode. However, downlink throughput throttling can be used to thermally mitigate increasing UE temperatures. As the temperature of the UE increases, the network can be notified to decrease the amount of downlink throughput, thus reducing the power levels and the temperature …
In-Page Search Over Text And Images In A Web Page, Matthew Sharifi
In-Page Search Over Text And Images In A Web Page, Matthew Sharifi
Defensive Publications Series
Browser applications provide users with the ability to search within a webpage. Such searches are limited to textual content of the page and do not take into account content depicted as images within the webpage. This disclosure describes techniques to enable search to be performed over the entire webpage, including image content. With user permission, images on a webpage are analyzed to determine presence of text within the image. If text is detected, OCR techniques are applied to recognize characters in the text. When the user enters a search query for searching within the page, the text content of the …
User Authentication Via Lip Reading, Horacio Hernan Moraldo
User Authentication Via Lip Reading, Horacio Hernan Moraldo
Defensive Publications Series
This disclosure describes techniques to prevent unauthorized access to information via a facial recognition interface. With user permission, a live video of the user attempting to access the information is obtained via a camera to perform facial recognition and determine the presence of the user. The user is instructed to dictate a randomly generated passphrase while facing the camera. Lip reading techniques are applied to validate the dictation of the randomly generated passphrase. Access to the information is granted only upon successful validation.
Auto-Recovery Of Micro-Services In A Large Distributed Appliance Model, Kshitij Tanay, Raghu Rajendra Arur, Taj Uddin
Auto-Recovery Of Micro-Services In A Large Distributed Appliance Model, Kshitij Tanay, Raghu Rajendra Arur, Taj Uddin
Defensive Publications Series
In large distributed application-based systems, an ensemble of co-existing micro-services perform individual functions and communicate with other micro-services. As these interactions and individual behaviors can be complex, it may be necessary to monitor micro-services for any abnormalities. Manual monitoring of all such micro-services, or establishing individual rule-based alerts for micro-services can be time-consuming. In addition to maintaining functionality of micro-services, it is important to prevent micro-services from becoming a performance bottleneck for other micro-services. In such situations, an alert may not suffice. For example, if a micro-service serves as a data store for all other services, a bottleneck at that …
Virtualized Intelligent Honeypot Agent, Plamen Nedeltchev, Mani Kesavan, Hugo Latapie, Enzo Fenoglio
Virtualized Intelligent Honeypot Agent, Plamen Nedeltchev, Mani Kesavan, Hugo Latapie, Enzo Fenoglio
Defensive Publications Series
A honeypot system is described that can expand to any attack surface as it learns and grows with the changing device landscape. The system also takes into account the human elements that originated the attack. By using adversarial training mechanisms, the system may be quickly trained to become a doppelganger and attract attacks. Moreover, a unique quantum cognitive framework provides a robust adaptivity to ever-changing attacker strategies. Virtualized intelligent honeypot agents may be introduced into the network, device, or server, to connect and share knowledge to facilitate federated learning for similar type of agents. The agents may also be operated …
Visual Approach For Secure Transfer Of User Credentials, Patrick Georgi
Visual Approach For Secure Transfer Of User Credentials, Patrick Georgi
Defensive Publications Series
This disclosure describes techniques to set up a user account, such as a single sign on account, on a new user device with the use of another device on which the user is already authenticated. The authenticated device displays a QR code that includes encrypted data that includes user credentials and optionally, other information such as wireless access point name and credentials. For example, the data is encrypted using a public key of the new device. The new device decrypts the data using its own private key. The described techniques eliminate the need for a user to enter credentials on …
Optimization Of Cloud Platforms, Johnson Manuel-Devadoss, David Stanford, Billie Ansell, Sunil John, Luis Bravo, Miguel Figueroa, Larone Gardner
Optimization Of Cloud Platforms, Johnson Manuel-Devadoss, David Stanford, Billie Ansell, Sunil John, Luis Bravo, Miguel Figueroa, Larone Gardner
Defensive Publications Series
Embodiments presented herein provide a mechanism for optimizing the cost of a cloud platform. A cognitive cloud platform is trained using performance counters to process unstructured data in order to identify underutilized cloud resources. The platform may also generate policies to shut down during idle times. Present embodiments may utilize techniques such as robotic process automation (RPA), straight through processing (STP), and/or natural language processing (NLP).
Intelligent Editing Assistant, Emmanuel Arriaga
Intelligent Editing Assistant, Emmanuel Arriaga
Defensive Publications Series
A technique is proposed for assisting writing or editing an online document supported by a cloud-content based platform. An editing assistant service provided by the cloud-content based platform determines a writing style designated for the online document. The editing assistant service determines a subset of content in the online document. Then, the editing assistant service determines whether the subset of the content in the online document is consistent with the writing style. Responsive to determining that the subset of the content in the online document is not consistent with the writing style, the editing assistant service predicts one or more …
Verifiably Secure Session Initiation Protocol Requests, Mark Barrasso
Verifiably Secure Session Initiation Protocol Requests, Mark Barrasso
Defensive Publications Series
Techniques are described for reducing the amount of spam and congestion on Session Initiation Protocol (SIP) devices and endpoints to significantly improve customer User Experience (UX). This may be packaged as a web Application Programming Interface (API) that provides an “anti-spam as a service” for other web-based clients.
Screen Brightness Adjustment Using Machine Learning, Sandro Feuz, Thomas Deselaers
Screen Brightness Adjustment Using Machine Learning, Sandro Feuz, Thomas Deselaers
Defensive Publications Series
Mobile devices, e.g., smartphones, tablets, etc., often include a light sensor that senses ambient light. Data from this sensor is used to set the screen brightness. During ordinary usage of such devices, the light sensor sometimes gets covered, e.g., by the user's fingers, which leads to a quick decrease in screen brightness. This makes for a poor user experience. This disclosure describes machine learning techniques that determine whether a detected change in ambient light is due to a true decrease in ambient light conditions or due to an occlusion of the light sensor. Screen brightness adjustments can be made on …
Radiofrequency Trends Aware Dynamic Bandwidth Selection, Santosh Kulkarni, Vishal Desai, Young Ii Choi, Pooya Monajemi
Radiofrequency Trends Aware Dynamic Bandwidth Selection, Santosh Kulkarni, Vishal Desai, Young Ii Choi, Pooya Monajemi
Defensive Publications Series
The current Dynamic Bandwidth Selection (DBS) algorithm only uses a snapshot of client metrics to make bandwidth recommendations for the remainder of the day. According to the present techniques, machine learning (ML) may be used to track/predict bias factors for DBS throughout the day.
Enhanced Call Quality Using User-Specific Voiceprint Model, Chilin Kuo, Hsinhu Wang, Ssu-Ting Tsai, Chin-Kuo Huang
Enhanced Call Quality Using User-Specific Voiceprint Model, Chilin Kuo, Hsinhu Wang, Ssu-Ting Tsai, Chin-Kuo Huang
Defensive Publications Series
The audio quality of a call conducted using a user device such as a phone can be unsatisfactory due to a variety of reasons. Such reasons include noisy surroundings at the location from which the user conducts a call, position of the phone near the user’s face, users that speak with a soft voice, presence of far-end echoes, etc. This disclosure describes techniques that use a user-specific voiceprint to home into the user’s speech and cut out surrounding disturbances from a voice call. The techniques are implemented with user permission to generate and use the voiceprint.
A Homotopy Method Of Path Analysis For The Purpose Of Computing, Lei Liu
A Homotopy Method Of Path Analysis For The Purpose Of Computing, Lei Liu
Defensive Publications Series
The present disclosure provides systems and methods for path analysis using a homotopy method, for the purpose of computing. In particular, the present disclosure unifies path representations with the standard construction and formalizes synthetics and semantics of the paths. In this way, the present disclosure enables efficient representation and manipulation of paths in well forms of all dimensions, including in higher dimensions. This is critical and significant to computing, for example, when path representations need to be transformed to each other during cloud resource and service migrations. Keywords associated with the present disclosure include: path analysis; homotopy; homotopy path analytics; …
Better Unpacking Binary Files Using Contextual Information, Armijn Hemel
Better Unpacking Binary Files Using Contextual Information, Armijn Hemel
Defensive Publications Series
To unpack firmware files, disk images, raw flash dumps, file systems or other archives various tools are available, that examine the contents of the file, find offsets of archives, compressed files, media files and so on, carve these from a larger file, decompress the carved files, and make the unpacked data available for recursive unpacking. Currently available tools treat all found files of a certain type the same (all PNG files are treated the same, all ZIP files are treated the same and so on), without taking the context in which they were found into account, which actually could matter …
Matching Entities On Multiple Platforms Using Graph Bijections, Long Nguyen, Zachary Frazier
Matching Entities On Multiple Platforms Using Graph Bijections, Long Nguyen, Zachary Frazier
Defensive Publications Series
Disclosed herein is an improved mechanism for matching entities on multiple platforms using graph bijections. The mechanism can identify two trees (e.g., a first tree and a second tree) within a given directed graph, where each tree has a root node corresponding to a particular entity, such as a reporting entity for advertisement metrics (e.g., impressions, views, etc.) or any other suitable entity. The mechanism can then identify the leaves associated with each tree. For each identified leaf, the mechanism can determine if there is advertising traffic or other suitable activity that connects the leaf with a leaf of another …
Identifying Abusive Videos Inserted In A Video, Filip Pavetic, Roman Vorushin
Identifying Abusive Videos Inserted In A Video, Filip Pavetic, Roman Vorushin
Defensive Publications Series
Disclosed herein is an improved mechanism for identifying abusive videos that have been inserted in a video. The mechanism can identify a video to be analyzed. The mechanism can then identify, for each frame of the video, a group of candidate locations within the video at which a two-dimensional video is likely to be embedded. The system can cluster the identified candidate locations and identify a subset of the candidate locations based on the clusters. The system can then identify a two-dimensional video embedded at one of the locations included in subset of candidate locations.
Design Of Neural Networks Based On Cost Estimation, Yair Movshovitz-Attias, Andrew Poon, Ariel Gordon, Elad Edwin Tzvi Eban
Design Of Neural Networks Based On Cost Estimation, Yair Movshovitz-Attias, Andrew Poon, Ariel Gordon, Elad Edwin Tzvi Eban
Defensive Publications Series
Certain automatic designs of neural networks not only minimize prediction error but also shrink or prune the network to reduce inference latency. Targeting inference latency directly is difficult; hence, FLOP-count is often used as a proxy for inference latency. However, FLOP-count is only loosely correlated with inference latency.
This disclosure describes techniques for direct computation or measurement of targeted costs such as inference latency, energy consumption, throughput, model size, etc. By integrating such targeted costs into design procedures, high performance neural networks of low inference latency, model size, and energy consumption can be obtained. The techniques find application in domains …
Zero-Touch Bootstrap Of A Network Connected Device, Owen Friel, Eliot Lear, Jerome Henry, Stephen Orr
Zero-Touch Bootstrap Of A Network Connected Device, Owen Friel, Eliot Lear, Jerome Henry, Stephen Orr
Defensive Publications Series
Techniques are presented herein that allow devices to automatically discover the correct enterprise wireless network to connect to, and securely onboard against that network, without manual provisioning of network information or credentials on the devices. This enables secure deployment of devices at scale on enterprise wireless networks. Minor enhancements to Institute of Electrical and Electronics Engineers (IEEE) 802.11 are described to enable this flow. Unlike Wi-Fi Alliance® Device Provisioning Protocol (DPP), the techniques presented herein are lightweight and do not include additional messaging overhead between the client device (STA) and Access Point (AP).
3d Face Messaging On A Watch, Vinay Venkataraman, Eric Caplain
3d Face Messaging On A Watch, Vinay Venkataraman, Eric Caplain
Defensive Publications Series
A system and method are disclosed that may send a video-like message to a receiver even when the receiver is offline. The system includes a user smart device incorporating an application, and a receiver smart device incorporating the similar application. The method captures a single 2D image of the user’s face using a camera incorporated in the communication device. The 2D image of the user’s face is then recreated as a 3D model. The message to be sent to a receiver is recorded as an audio message and sent along with the 3D model. At the receiver end the audio …
Software For Guided Chassis Adjustment, Verena Schwaiger
Software For Guided Chassis Adjustment, Verena Schwaiger
Defensive Publications Series
No abstract provided.
Shelf Label Flag, Verena Schwaiger
Shelf Label Flag, Verena Schwaiger
Defensive Publications Series
No abstract provided.
Natural Hmi For Efficient Driving Through Plants In The Car, Verena Schwaiger
Natural Hmi For Efficient Driving Through Plants In The Car, Verena Schwaiger
Defensive Publications Series
No abstract provided.
Method And System For Compensating Yaw Rate Change By Load Change Before Or During Cornering, Verena Schwaiger
Method And System For Compensating Yaw Rate Change By Load Change Before Or During Cornering, Verena Schwaiger
Defensive Publications Series
No abstract provided.
Extension Of The Current Flow Diagnosis Of The Dc-Charging Interface, Verena Schwaiger
Extension Of The Current Flow Diagnosis Of The Dc-Charging Interface, Verena Schwaiger
Defensive Publications Series
No abstract provided.
Cyclone Separators For Screw Systems, Verena Schwaiger
Cyclone Separators For Screw Systems, Verena Schwaiger
Defensive Publications Series
No abstract provided.
Bidirectional Dc Wallbox With Dc-Bus Interface And Integrated Bidirectional Energy Meter, Verena Schwaiger
Bidirectional Dc Wallbox With Dc-Bus Interface And Integrated Bidirectional Energy Meter, Verena Schwaiger
Defensive Publications Series
No abstract provided.
Detecting The Direction Of Ambient Light Using An Array Of Photodiodes, Yi Zhang, Seobin Jung, Kuan-Lin Chen
Detecting The Direction Of Ambient Light Using An Array Of Photodiodes, Yi Zhang, Seobin Jung, Kuan-Lin Chen
Defensive Publications Series
So as to gain context, mobile devices, e.g., phones, tablets, laptops, etc., capture information about their surroundings, including sounds, lights, temperature, etc. An important piece of information is the direction of nearby light sources. Currently available ambient light sensors provide illuminance information so that the display can adjust its brightness accordingly. However, there is no directional information provided.
This disclosure describes techniques that use a planar array of ambient light sensors with asymmetric masks to determine the angle of incident light.
Enabling Instantaneous Switch From Default Multicast Distribution Tree To Data Multicast Distribution Tree For Mobility And Multihoming In Ethernet Virtual Private Network Tenant Routed Multicast, Raunak Banthia, Manoj Pandey, Kesavan Thiruvenkatasamy, Saif Mohammed, Vikas Ramesh Kamath
Enabling Instantaneous Switch From Default Multicast Distribution Tree To Data Multicast Distribution Tree For Mobility And Multihoming In Ethernet Virtual Private Network Tenant Routed Multicast, Raunak Banthia, Manoj Pandey, Kesavan Thiruvenkatasamy, Saif Mohammed, Vikas Ramesh Kamath
Defensive Publications Series
Techniques are described for providing an optimized way of switching from a default Multicast Distribution Tree (MDT) to a data MDT to handle two key use-cases in Ethernet Virtual Private Network (EVPN) / Tenant Routed Multicast (TRM). The first key use-case is an Ethernet Segment Identifier (ESI) failover case over a multihomed network in an EVPN fabric. The second use-case is host mobility.
A Virtual Call Assistant, Anonymous
A Virtual Call Assistant, Anonymous
Defensive Publications Series
A system for adding a virtual assistant as a user to a call is disclosed. The virtual assistant (implemented as a web Real Time Communication (webRTC) client) can be added by a manual selection on a Graphical User Interface (GUI) of a calling application (app) when a call is in progress between users. The virtual assistant responds to queries of the users. The users may ask queries from the virtual assistant using a trigger phrase, such as “Hi assistant!”. The virtual assistant implements speech recognition algorithms and searches for requisite information in a knowledge graph over the Internet and answers …