Open Access. Powered by Scholars. Published by Universities.®
- Publication Year
- File Type
Articles 5131 - 5160 of 11865
Full-Text Articles in Entire DC Network
Vectorization-Aware Loop Unrolling, Ishika Ahuja
Vectorization-Aware Loop Unrolling, Ishika Ahuja
Defensive Publications Series
The present disclosure discloses a computer implemented method and apparatus of vectorization aware loop unrolling having sequence of operation which includes the following steps : reading of a loop by the neural network; upon reading a loop, identifying the vectorizable instructions from a loop by a neural network, extracting essential features from the identified vectorizable instructions, estimating the vectorization window lengths using the decision tree algorithm based on the features of the vectorizable instruction and also estimating the profitability based on the profitability estimated in previous attempts using the neural network, computing the potential of the Straight-Line-Code Vectorization (SLP) as …
Product Search Refinements Based On Inventory Data, Aashiya Begum, Kristoffer Lidstrom, Nitzan Ishai
Product Search Refinements Based On Inventory Data, Aashiya Begum, Kristoffer Lidstrom, Nitzan Ishai
Defensive Publications Series
Store navigation queries can often be broad in nature, with user intent ranging from navigational (getting directions to the store or a link to the store website) to evaluating a store for a visit based on store vibe, inventory selection, product availability, etc. Search results for a local store can be unhelpful if the inventory for the store is unavailable to the search engine. This disclosure describes techniques that reduce or eliminate hurdles in online shopping journeys by displaying product categories and alternative stores that are relevant to a user query. The techniques enable users to continue or start their …
Identifying Risks And Improving Test Coverage In Software Binaries, Shu-Wei Cheng, Kamakshi Kodur, Leila Al-Muhtasib, Christopher Charles Burns, Patricia Legaspi, Bradley Andrews, Andrew Huang
Identifying Risks And Improving Test Coverage In Software Binaries, Shu-Wei Cheng, Kamakshi Kodur, Leila Al-Muhtasib, Christopher Charles Burns, Patricia Legaspi, Bradley Andrews, Andrew Huang
Defensive Publications Series
This disclosure describes techniques to verify test coverage within codebases that are subject to frequent modifications, e.g., codebases under development or maintenance. Codebase areas touched by an added feature are identified and focused upon during testing. Test coverage, both automated and manual, of the changed areas of the code is assessed. Areas requiring additional testing are highlighted and their risk levels visualized with a heatmap. The techniques include services that perform on-device profiling with on-demand start-and-stop; generate heatmaps of code modifications within binaries; list methods and generated test cases; generate mappings from test targets to executed client methods to executed …
Delivering Advertisements With Unmanned Aerial Vehicles (Uavs), N/A
Delivering Advertisements With Unmanned Aerial Vehicles (Uavs), N/A
Defensive Publications Series
This disclosure describes techniques to employ UAVs to deliver personalized advertisements selected based on the preferences of user groups at locations in the vicinity of the UAV. The UAV can deliver the personalized advertisements by dynamically moving between locations as appropriate to optimize the return on investment (RoI) for the advertisements. Interests of user groups at different locations in the vicinity of the UAV are determined and matching advertisements are selected. The combinations of location and advertisement are ranked to determine the UAV itinerary. The itinerary is updated dynamically based on changes to the user groups/ interests.
Cache For Building Piecemeal S2 Geometric Heatmaps, Becker Bradley Awqatty, Michael Sprague
Cache For Building Piecemeal S2 Geometric Heatmaps, Becker Bradley Awqatty, Michael Sprague
Defensive Publications Series
Heatmap visualizers traditionally rely on the assembly of a complete dataset prior to the creation of a heatmap. However, acquiring the complete dataset can be impractical when dealing with dynamic, ad-hoc, or diverse data sources. This disclosure describes techniques of building heatmaps using S2 cell-mapping, where interconnected cells overlay a globe in a hierarchical structure. The techniques involve identifying cells visible to a user; retrieving raw data pertaining to the identified cells; and constructing a heatmap layer by layer. The layers form a heatmap pyramid, with each level summarizing the data below it. At any time, data relevant to the …
Seamless Ink-Graphics Interleaving And Authoring Via Structural Decompilation Of Smart Documents, Daniel Kiecza
Seamless Ink-Graphics Interleaving And Authoring Via Structural Decompilation Of Smart Documents, Daniel Kiecza
Defensive Publications Series
This disclosure describes a digital educational environment that provides a frictionless editing experience for students and enables them to focus on their learning and can also be used in other settings. Traditional multimedia content is seamlessly blended with handwritten content and semantically interpreted by content type to provide scaffolded learning. Interactive learning experiences in text, charts, tables, shapes, mathematics, games, diagrams, graphs, images, drawings, data visualization, etc. are supported in a multi-representational form. A student can solve an equation using either stylus/ink or math equations rendered from ink strokes, and the solution can be automatically and parallelly tested for correctness. …
Obtaining Interest Group Identifiers From Url Address Bar, Wei Huang, Zhenyu Liu, Michael Kieber
Obtaining Interest Group Identifiers From Url Address Bar, Wei Huang, Zhenyu Liu, Michael Kieber
Defensive Publications Series
API scripts can be used to send information identifying user’s interests from their devices to entities, e.g., content publishers for use in customizing content for the users. However, in some situations, e.g., when a user enters a search query in an address bar of a web browser, the scripts may not be executed. Other techniques can be used to send the information identifying user’s interests, such as macros, HTTP headers, and/or cookies.
Dynamic Resource Requirement And Provisioning Timing Prediction For Automatically Scaling Computing Resources, Dhandapani Shanmugam
Dynamic Resource Requirement And Provisioning Timing Prediction For Automatically Scaling Computing Resources, Dhandapani Shanmugam
Defensive Publications Series
Resource requirements of cloud-hosted applications can change dynamically during application execution. Such dynamic changes are usually handled by automatic scaling mechanisms offered by cloud platforms. The reactive nature of auto scaling introduces a lag between when the need to provision additional resources to an application is detected and the actual allocation of the additional resources. Over-provisioning and manual allocation approaches make suboptimal use of resources. This disclosure describes a machine learning model that utilizes application characteristics and infrastructure attributes to predict the time at which resource requirements are likely to change and to generate the prediction timely such that proactive …
Optimizing Data Access Using Symbolic Links And Machine Learning, Puneet Mahajan
Optimizing Data Access Using Symbolic Links And Machine Learning, Puneet Mahajan
Defensive Publications Series
Traditional data management practices replicate data across data centers and regions to ensure availability and performance. Such data replication can result in high operational costs, complex data management, inefficient resource utilization, etc. This disclosure describes techniques that address the inefficiencies and complexities of existing cloud storage and data access mechanisms by leveraging the simplicity and efficiency of symbolic links for data tables and by dynamically optimizing data access based on usage patterns discovered using machine learning. Unnecessary data replication is reduced, leading to simplified data management, efficient resource utilization, and lower operational costs. Data access in distributed environments is made …
Enabling Subscription Awareness For Card Based Payment Ecosystem, Burnnet Tai Lung Or Visa, Richard Stopic Visa, Warren Everard Visa, Hollie Buchanan Visa
Enabling Subscription Awareness For Card Based Payment Ecosystem, Burnnet Tai Lung Or Visa, Richard Stopic Visa, Warren Everard Visa, Hollie Buchanan Visa
Defensive Publications Series
Present disclosure introduces a concept of subscription awareness by requesting merchants to provide subscription details to the network token service provider with cardholder provided card credential. With such subscription information provided, token service provider may tie such information to the token being generated which in turn ties to the card credential and the subscription itself. By having this subscription information flowing in the payment processing networks, a path may be provided where subscription information to flow in the card payment ecosystem between seller, acquirer, network, issuer and ultimately, card holders. Furthermore, with subscription information held along with token within token …
Transnet: Transaction Networks For Tabular Data, Minghua Xu, Huiyuan Chen, Menghai Pan, Yuzhong Chen, Yingtong Dou, Mahashweta Das
Transnet: Transaction Networks For Tabular Data, Minghua Xu, Huiyuan Chen, Menghai Pan, Yuzhong Chen, Yingtong Dou, Mahashweta Das
Defensive Publications Series
The present disclosure relates to a technique for implementing data loading in hierarchical pipeline for maximum utilization of computer resources (like GPU, etc.) and to obtain effective prediction performance from the numerical and categorical features. The technique involves dividing the data into chunks of data, batching them to form tensors and loading the batched data to the framework to be executed on the GPU. The framework generates embeddings for the numerical and categorical features identified from the batched data and determines the interactions between the generated embeddings to provide a predictive output.
Solid State Forms Of (E)-2-Isopropyl-5-Styrylbenzene-1,3-Diol And Processes For Preparation Thereof, Anonymous
Solid State Forms Of (E)-2-Isopropyl-5-Styrylbenzene-1,3-Diol And Processes For Preparation Thereof, Anonymous
Defensive Publications Series
The present invention provides solid state forms of (E)-2-isopropyl-5-styrylbenzene-1,3-diol and processes for preparation thereof.
Virtual Quadcopter, Colin Hilton
Virtual Quadcopter, Colin Hilton
Defensive Publications Series
A multicopter comprises an arrangement of spars that normally makes a four-pronged set of cantilevers around a centre body, each supporting a single motor and propeller. The layout described here extends each cantilever to support a pair of close-coupled motors that rotate in the same direction, contrary to received practise. This creates a form of virtual quadcopter that provides failure redundancy yet improved efficiency in comparison to an X-8 layout, whilst still retaining a footprint suited to human carriage.
Using An Attestation Key To Enhance Smart Device Setup Security, Jingjiang Li
Using An Attestation Key To Enhance Smart Device Setup Security, Jingjiang Li
Defensive Publications Series
Smart appliances and devices such as thermostats, lights, locks, cameras, doorbells, televisions, refrigerators, ovens, dishwashers, and robotic vacuum cleaners can be connected to the internet. A user may connect to the smart devices using one or more of their computing devices such as a smart phone, a laptop computer, and/or a tablet computer. Once connected to the smart devices which involves the use of one or more applications, the user may individually control each smart device remotely. The user of the smart devices can connect to, control, and monitor their smart devices from within the location of the devices (e.g., …
Highlighting Screen Share Content Based On Audio, Jieyu Tan, Soya Li, Xinxing Zheng
Highlighting Screen Share Content Based On Audio, Jieyu Tan, Soya Li, Xinxing Zheng
Defensive Publications Series
Techniques described herein combine artificial intelligence (AI)-based audio transcription (speech-to-text) with AI-based sharing content perception (frame-to-text). In particular, techniques provide for identifying spoken keywords/objects/areas during a communication session and highlighting the corresponding keywords/objects/areas on content being shared during the communication session. The highlighting may be performed automatically, dynamically, and instantaneously to point out features in the shared content that are being described or mentioned by a speaker.
Industrial Network Segmentation Model With Sub-Zone And Sub-Conduit Support, Elango Ganesan, Flemming S Andreasen, Michael Freed, Ruben E Escolero
Industrial Network Segmentation Model With Sub-Zone And Sub-Conduit Support, Elango Ganesan, Flemming S Andreasen, Michael Freed, Ruben E Escolero
Defensive Publications Series
Techniques described herein provide for models for industrial networks that support the concepts of sub-zones and sub-conduits. The models formalize sub-zones and sub-conduits and define a hierarchy for the sub-zones and sub-conduits. The models include well-defined rules for the creation of sub-zones and placement of assets into the sub-zones based on the hierarchy structure and the creation of sub-conduits between zones and sub-zones with a hierarchical set of communication restrictions based on ancestor conduits.
Techniques To Facilitate Multi-Access Point Coordination (Mapc) In A Wireless Local Area Network (Wlan) Environment, Malcolm Smith, Brian Hart
Techniques To Facilitate Multi-Access Point Coordination (Mapc) In A Wireless Local Area Network (Wlan) Environment, Malcolm Smith, Brian Hart
Defensive Publications Series
Standards organizations involved in the development of next generation wireless local area network (WLAN) standards, such as the Institute of Electrical and Electronics Engineers (IEEE) 802.11 and 802.11bn standards, and amendments thereof, are discussing the possibility of WLAN operation involving a shared transmit opportunity (TXOP) under multi-access point (AP) coordination (MAPC) control in which a leader AP can initiate each of a TXOP with a follower-to-leader information exchange and then a notional multi-AP trigger frame. Current proposals for such operation do not define contents of the trigger for the range of MAPC mechanisms, such as spatial re-use, Time-Division Multiple Access …
Iterator Facade Layer With Multiple Pointers To Facilitate Access To Columnar Data, Prashant Mishra, Prateek Jain, Shevek .
Iterator Facade Layer With Multiple Pointers To Facilitate Access To Columnar Data, Prashant Mishra, Prateek Jain, Shevek .
Defensive Publications Series
For faster processing and higher efficiency, columnar data storage is preferred for many applications. However, data stored in columnar format may be required in row-oriented format in certain use cases. For large analytical databases, transforming data from column to row format and storing the transformed data can be expensive. This disclosure describes techniques to efficiently access columnar data in a database without transforming the data into row-oriented format. An iterator facade layer with multiple pointers to columnar data is used to present data to applications that require data in a row-oriented format. Columnar data is retrieved in batches that are …
Rendering Images Based On Metadata Related To Ambient Light Conditions, Sianyi Huang, Jongseo Lee
Rendering Images Based On Metadata Related To Ambient Light Conditions, Sianyi Huang, Jongseo Lee
Defensive Publications Series
This disclosure describes techniques to render an image based on metadata that is related to ambient light conditions of a reference display environment for the image. Ambient light information is detected in the reference environment in which an image is displayed by a reference device. The ambient light information, as well as display brightness and gamma value settings of the reference device, are stored as metadata with the image. When decoding the image for display on a target device, this metadata is used to automatically adjust display brightness and/or gamma value of the target device to match the reference display …
Natural Language Interface Enhanced Camera Using Large Language Models, Kun Wang, Wei Hong, Li Wei
Natural Language Interface Enhanced Camera Using Large Language Models, Kun Wang, Wei Hong, Li Wei
Defensive Publications Series
Modern cameras, including smartphone cameras, are powerful and provide a large set of features that enable users to capture photos/videos in a variety of conditions and per their aesthetic preferences. However, such features are accessed via menus and complex user interfaces that are difficult for a user to navigate. This disclosure describes techniques that utilize a large language model (LLM) to provide a natural language interface to enhance traditional camera functionality. By utilizing an LLM, the user interaction is made more intuitive as users can communicate with their camera using natural language, reducing or eliminating the reliance on complex interfaces, …
Configurable Sampling Curves For Data Logging And Telemetry, Ashok Chandwaney, Cindy Zhou, Branden Archer
Configurable Sampling Curves For Data Logging And Telemetry, Ashok Chandwaney, Cindy Zhou, Branden Archer
Defensive Publications Series
Telemetry can be added to measure the function of application and device features. As feature stability increases, telemetry becomes less necessary. However, ramping down telemetry requires engineering effort and not all devices may receive configuration updates to reduce telemetry. This disclosure describes techniques that enable software or hardware developers to specify a log-sampling curve that automatically ramps down telemetry over time. The techniques enable developers to obtain telemetry data in a manner that optimizes data consumption and footprint. The techniques advantageously result in a long-term reduction in the volume of telemetry data without an abrupt drop-off in such data. Close …
Eliminating Bluetooth Audio Glitches Caused By Clock Drift, Dc Xu
Eliminating Bluetooth Audio Glitches Caused By Clock Drift, Dc Xu
Defensive Publications Series
When a device has Bluetooth host and controller clocks that reside on different clock domains, the clocks may drift apart over time (not be synchronized). This can cause communication between the two domains to experience dropped or missed packets, leading to problems such as audio drops, glitches, etc. that affect any protocols that use software codecs which rely on precise timing and clock synchronization. This disclosure describes techniques to reduce the impact of clock drift. These include use of a trigger by the Bluetooth controller, e.g., a general purpose input/output (GPIO) trigger or a hardware interrupt from a host controller …
Enhancing Detection Of Small Objects In Images By Using Face Regions, Kun Wang, Wei Hong, Vino Huang, Tianyi Bi, Kerwin Huang
Enhancing Detection Of Small Objects In Images By Using Face Regions, Kun Wang, Wei Hong, Vino Huang, Tianyi Bi, Kerwin Huang
Defensive Publications Series
Current object detection models face difficulties in detecting small but important objects within large frames, posing difficulties in applications such as security cameras or camera timers for selfies or group photographs that are based on detection of hand gestures. This disclosure describes techniques that utilize additional signals from image frames to improve the detection rate and location accuracy of small objects within the frame. For example, face detection can be performed and detected faces can be used as anchors to detect nearby hand gestures that are relatively small. Faces can be clustered and regions around the faces cropped and enlarged …
Transforming Code To Anonymized Abstract Syntax Trees For Ai/Ml Model Training, N/A
Transforming Code To Anonymized Abstract Syntax Trees For Ai/Ml Model Training, N/A
Defensive Publications Series
A collection of code bases that include source-destination pairs of code, translated from a source environment to a destination environment can be highly valuable for training artificial intelligence (AI) or machine learning (ML) models. However, it is possible that a code base includes private or sensitive information such as variable names specific to a particular party, which makes it infeasible for such use. This disclosure describes techniques to automatically remove sensitive information from code to make the code amenable for use as training data for machine learning (ML) or artificial intelligence (AI) models. Source-destination pairs of translated code are transformed …
Compiler Categorization And Llm-Based Code Translation Via Few-Shot Learning, N/A
Compiler Categorization And Llm-Based Code Translation Via Few-Shot Learning, N/A
Defensive Publications Series
Heterogeneous database or code migration entails the conversion of schema and/or code from a source language to a target language. Current tools only partially address the database migration problem and full migration requires labor and time intensive manual developer intervention. This disclosure describes techniques to reduce or eliminate the human intervention required for code/schema migration by automatically identifying and categorizing recurring instances of objects or code that are resistant to automatic translation, referred to as gaps in translation. A large language model (LLM) is trained, e.g., using few-shot learning, to translate such gaps in translation using developer-provided training examples illustrating …
Automatic Categorization Of Log Files And Incident Reports, Shrey Nagpal, Souvik Paul, Shubham Sharma, Prerna Kakkar
Automatic Categorization Of Log Files And Incident Reports, Shrey Nagpal, Souvik Paul, Shubham Sharma, Prerna Kakkar
Defensive Publications Series
Log files and reports describing outages and bugs include free-form text fields that include information about system outages or bugs in a computer system. However, identifying patterns such as common types of bugs, affected users, critical components or binaries, etc. manually from such documents can be tedious. This disclosure describes techniques to automatically categorize documents in multiple relevant categories based on the key features extracted from free-form text information in the documents. The extracted features are subjected to dimensionality reduction. Clustering techniques are applied to automatically identify clusters that are then used to generate visualizations. In the context of documents …
Automating Task Scheduling And Resource Allocation For Shared Computing Resources, Amy Shen, John Schuett, Christine Kingsley
Automating Task Scheduling And Resource Allocation For Shared Computing Resources, Amy Shen, John Schuett, Christine Kingsley
Defensive Publications Series
Optimally scheduling shared computing resources for a variety of tasks with different task-specific resource requests is a difficult problem. Scheduling computational resources is labor intensive, time consuming, error prone, and manual allocation can result in suboptimal use of computing resources. This disclosure describes techniques that train and utilize a machine learning model to automate the task scheduling and resource allocation for the use of shared computing resources. The model is trained on historical records of task requests and corresponding scheduling, allocations, runtime outcomes. From the training data, a score can be computed for each request, indicating the extent to which …
Architecture For Converting Natural Language To Sql Queries, Kanchana Patlolla, Steve Walker, Carlos Augusto
Architecture For Converting Natural Language To Sql Queries, Kanchana Patlolla, Steve Walker, Carlos Augusto
Defensive Publications Series
The ability to automatically convert natural language (NL) queries to structured query language (SQL) queries (NL2SQL) can improve the accessibility of databases to end users. This disclosure describes a standardized NL2SQL architecture pattern (including building blocks) for querying structured data using natural language. The probabilistic nature of large language models (LLMs) is leveraged to generate SQL queries by understanding the metadata of structured databases. The generated SQL queries match the dialect of the database, including its version and package, and are validated and tested for accuracy.
Lead Enrichment Automation, N/A
Lead Enrichment Automation, N/A
Defensive Publications Series
Lead enrichment - enhancing or augmenting existing marketing lead data with additional information - helps improve targeting and personalization in marketing campaigns. Lead enrichment involves augmenting data using third-party databases to enrich existing lead profiles, which when performed manually has a long turnaround time due to the requirement of substantial manual effort. This disclosure describes techniques that automate the lead enrichment process by extracting data from third-party databases by querying such databases via appropriate application programming interface (API) calls that specify the available lead information. The databases can be accessed based on user-specified priority and complete/partial enrichment achieved by use …
Image Encryption Using Physical Properties Of Sensor, Jason Mayes
Image Encryption Using Physical Properties Of Sensor, Jason Mayes
Defensive Publications Series
In digital cameras, light passes through a lens and is detected by an image sensor. The raw frame(s) captured by the sensor is passed through a software stack to obtain an image or video. Any encryption is performed in the software stack. If the software stack is compromised prior to encryption, malicious actors can remotely access sensor data. Encryption and decryption also introduces additional computational load on both transmitter and receiver devices, and can lead to delays in the captured image/ video being available for viewing. This disclosure describes techniques that utilize the physical properties of glass that has been …