Digital Commons Network^™

Techniques To Expand Grammar Pattern Matching, Amit Singhal, Michael Ng

Defensive Publications Series

Language understanding is an important aspect of supporting spoken queries on a device. Intensive ML models are unsuitable on resource-constrained devices and in such cases, pattern matching grammar can be utilized. However, pattern matching can have low recall and may be limited in understanding user queries. This disclosure describes the use of normalization, optional/stopwords, synonym clusters, and argument expansion to generate more expansive grammar with limited training data and without the use of machine learning. The use of such a grammar can improve recall while maintaining high precision. The techniques are suitable for any context where grammar patterns are used …

Method And System For Determining Sustainability Information In Authorisation Message, Hyeonmin Na, Sajjad Rizvi, Balakrishna Chittari

Defensive Publications Series

The present disclosure relates to method and system for calculating sustainability information in authorisation message based on transaction data. The present invention uses payment rails, sourcing for delivering sustainability information. The method also comprises sourcing and providing payment network entities and consumers the information on carbon intensity of a purchase. The present disclosure provides a solution for calculating the sustainability score of the registered Permanent Account Number (PAN) using transaction history, where transaction history is used to establish direct connectivity between merchants and issuers for the purpose of processing card transactions.

Autonomous Collection Detection And Remediation Decisions Based On Local Models And Locally Sourced Data, Dave Zacks, Jane (Zizhen) Gao, Nagendra Kumar Nainar, Carlos M. Pignataro, Dmitry Goloubew

Defensive Publications Series

In context of distributed monitoring and anomaly detection, when a networking device performs anomaly detection based on local data, such as when a remote controller is not reachable during network convergence or other network issues. Anomaly relevance improves if telemetry data used for anomaly detection comes not only from a local device, but also from the device's immediate surroundings (e.g., physical neighbors, protocol peers, redundancy units, etc.). Presented herein are techniques through which a device can reach its own and nearby telemetry sources in a manner that may follow an effective network topology and configuration. Thus, techniques herein may enable …

Navigating User Interfaces On A Mobile Computing Device, Yun Sun Lee, Luke Brantingham, Kavinaath Murugan, Mike Bentz, Aaditya Kandibanda

Defensive Publications Series

This publication describes systems and techniques for navigating user interfaces by moving or tapping a computing device (e.g., a smart phone, a computerized watch, computerized eyewear, laptop computer, or any other wearable or non-wearable mobile computing device). The computing device may include motion sensors (e.g., gyroscope, accelerometer, magnetometer, etc.) and/or pressure sensors (e.g., strain gauges, barometric sensors, switches, pressure plates, capacitive sensors, resistive sensors, etc.). Such sensors may detect tapping on the computing device and/or rotation of the computing device and provide the sensor data to the computing device. Using the data generated by one or more of these sensors, …

Algorithm To Estimate Suspended Metal Particles In Recirculation Agent Delivery System, Hp Inc

Defensive Publications Series

It is common for printers that use challenging agent or inks to include a Recirculation System through the

printhead. This allows for the agent/ink to mix both in the Agent Delivery System and the Printhead.

This method is especially useful in 3D-metals printers, where printhead reliability is crucial and refreshing

the printhead with new agent will increase the printhead life and achieve a better isothermal control in the

firing chamber.

One risk of recirculating through the printhead is that particles in suspension in the air of the print zone

might be ingested through the nozzles and end up in the …

Detecting Data Lineage Using Variable Block Deduplication, Sam Silverberg

Defensive Publications Series

With burgeoning data volumes, the need for accurate data cataloging, lineage, and governance is gaining importance. Traditional automation for data context discovery requires manual tagging or scanning of data using a parser to understand the content, which can require knowledge of the content type and can be error-prone, time-consuming, and expensive. Further, traditional data context discovery doesn’t track lineage between disparate data types, and cannot index binary data. This disclosure describes efficient techniques to index and map similarity across multiple datasets to determine data lineage. The techniques do not need or use the context of the underlying data or concepts …

Purge Process To Reduce Ink Spills, Hp Inc

Defensive Publications Series

In some printers, during the start‐up, the Ink Delivery System is purged by using a gas purger

(instead of a printhead) fluidically connected to the fluid interconnect on the carriage, which

allows to evacuate air or any other gas that may be present in the Ink Delivery System.

The gas in the Ink Delivery System is purged through the gas purger in response to a fluid being

pumped into the Ink Delivery System. The ink pressure gradually increases while the ink moves

upstream to the fluid interconnect. Once the ink reaches the purgers and the foam collapses,

the ink pressure …

Combined Head Gestures For Improved User Interaction, Lei Shi, Aobo Zhou

Defensive Publications Series

Fine-grained gestures such as eye gaze, facial expressions, etc., can be useful as input mechanisms for smart devices. However, single gesture inputs such as eye gaze are inefficient, and it is difficult to perform complex operations with such inputs. This disclosure describes techniques to fuse multiple head gestures, e.g., head, eye, mouth, or eyebrow movement, to provide superior user interaction. Head gestures are classified as analog (e.g., eye movements, which provide continuous input) or binary (e.g., frowns, which indicate one-time operation). An analog gesture is fused with multiple binary gestures to improve efficiency. For example, to move an object, the …

Alert Generation For Cyclically Varying Metrics, Joseph Turner

Defensive Publications Series

This disclosure describes techniques for monitoring computer system metrics that exhibit cyclical (periodic) behavior. Per techniques of this disclosure, function-based thresholds are utilized for metric monitoring and generation of system alerts. The function-based thresholds are time-varying and are aligned with the troughs and peaks of the underlying metric being monitored. Metric measurements are obtained for a particular period of time. A function such as a Discrete Cosine Transform (DCT) is fitted to the obtained metric data. A threshold is set for the metric based on the fitted function at a fixed offset from the function derived value. Function-based thresholding can …

Reliable Wireless Solution For Time-Critical Communications In Smart Utility Environments, Pascal Thubert, Huimin She, Li Zhao, Patrick Wetterwald

Defensive Publications Series

The Generic Object Oriented Substation Event (GOOSE) protocol is a communication model that is defined by the International Electrotechnical Commission (IEC) 61850 standard which supports the sharing of time-critical information between Intelligent Electronic Devices (IEDs) within a substation. Due to strict requirements, GOOSE demands a fast, reliable, and deterministic network, which today is based on Ethernet networking technologies. Customers desire an alternative solution that is based on wireless radio technologies such as, for example, mesh-based networks. However, for a variety of reasons, it can be challenging to implement GOOSE over a wireless medium. Techniques are presented herein that support a …

Reliable Normalized Customer Reviews, Alexander Ventura, Andy Lavery

Defensive Publications Series

Online reviews suffer from certain reliability problems such as lack of reviewer objectivity, reviewer bias, lack of domain familiarity on the part of the reviewer, etc. This disclosure describes techniques to improve the reliability of review scores obtained from online review facilities. A set of meta-parameters, referred to as review reliability parameters (RRP), is defined and associated with the underlying numerical rating of a review. RRPs are inputs to a machine learning model that derives a weighted review. The review of a reviewer with good RRPs is weighted more than the review of a reviewer with poor RRPs. RRPs can …

Leveraging Rich Call Data To Enhance Customer Experience, Kaustubh Inamdar, Gonzalo Salgueiro, Sreekanth Narayanan

Defensive Publications Series

Secure Telephony Identity Revisited (STIR) is an effort currently being utilized to provide cryptographic assurance of caller Identity in an effort to combat robocalls and impersonation attacks. The STIR framework can be used to transport additional claims during the course of a call. These additional claims, known as Rich Call Data (RCD), may be leveraged to embed useful, context-driven information. In particular, techniques presented herein provide an overarching framework through which companies that have a technical support division can advertise a set of RCD keys that would allow callers to avoid having to navigate interactive voice response (IVR) menus, specify …

Naturalistic Routing Using Inverse Reinforcement Learning, Renaud Hartert, Matt Barnes, Matthew Abueg

Defensive Publications Series

This disclosure describes techniques, referred to as naturalistic routing (NR), that improve the quality of routes found by map applications by learning from users’ real-world navigation actions, accessed with user permission. The techniques leverage the principle that users, in the aggregate, tend to travel on optimal routes to reach their destinations. A machine learning model is trained using inverse reinforcement learning and provides routes that are optimal by the users’ definition of optimality, as determined from a dataset of navigation actions.

Acoustics Project Report, Hp Inc

Defensive Publications Series

This project focuses on diagnosis of problems in a printer from captured acoustic

signals. Previous researchers already successfully designed a feature generator to

encode the acoustic signals into 6 dimensional features. And they also trained

classifiers on artificial data to predict if a printer is in a good or bad status. The

final target is to make a diagnosis based on the acoustic data, such as pre-alert.

This will help engineers to fix the problems of the printers before customers

making service calls.

Dependent Request Inference And Execution In Integration Tests, Patrick Blesi

Defensive Publications Series

The test setup phase of an automated software test for an application programming interface (API) typically requires the creation of many dependent objects. Current techniques for test setup typically require a test writer to explicitly define dependencies between resources, which entails substantial manual effort to build the setup necessary to test an API request. This disclosure describes techniques to automatically assemble a dependency graph and build the setup and cleanup necessary to test an API request. An API service definition is analyzed to enumerate all remote procedure calls (RPCs) and infer the fields and RPCs required to successfully execute a …

Key-Value Store Using A Network Packet Filter Situated In The Os Kernel, Andrew C. Jackson

Defensive Publications Series

When thousands of machines coordinate a small piece of information, a network key-value store can struggle to keep up with the query load. An underlying cause of the capacity bottleneck is that the key-value store is usually held in user space, not kernel space. This disclosure leverages the run-in-kernel capabilities of a network packet filter to deliver very fast responses to key-value (hash map) requests. The network packet filter is capable of communicating with hash-map type memory accesses and can rewrite and redirect packets. When assembled this way, a packet with the query key can be returned to the requester …

Identifying Software Projects That Have Matching Static Or Dynamic Characteristics, Andy Lavery, Eli Daiches

Defensive Publications Series

This disclosure describes techniques to analyze collections of software projects using static and dynamic features. Software engineers or operators can query the outcomes of these analyses to find cohorts of design patterns that are relevant for the task at hand in their own projects, thus helping them identify applications that meet the criteria of interest. The techniques can also be applied to generate alerts based on evolution of the matching projects, suggest code completion, and to recommend relevant example code or learning resources. Analysis of code can be performed internally on codebases controlled by a single business and/or on permitted …

Secure Wireless Localization Using Trusted Execution Environments, Avinash Kalyanaraman, Eric Voit

Defensive Publications Series

Through wireless localization, the position of a device (user) can be computed in a world coordinate system using wireless signals. Wireless localization systems for estimating device/user location typically involve fusing measurement data obtained from multiple access points (APs) by compute end-point, such as a cloud/edge-node. Such systems, however, are prone to attacks that can be launched against a compute end-point. To address such issues, techniques herein involve running wireless localization algorithms on a new paradigm of computing involving secure enclave technology through which localization algorithms can be protected from tampering. As a result, applications that may be vulnerable to location …

Fluid Separator, Bernd Lienau

Defensive Publications Series

This disclosure relates to a device for the separation of a mixture of a liquid and a gas. Disclosed is also a method relating thereto.

System And Method For Providing Service For A Search Request, Girish Balakrishna Hegde, Priyadarshini Sarode, Rahul Powar

Defensive Publications Series

The present disclosure provides a system and a method for providing service for a search request. The proposed system receives the search request entered by a user on a browser. The proposed system adds custom header to the search request to allow index-based extraction of address of the search request. Further, the proposed system extracts the address based on the length and hashes the address. The proposed system looks into a predefined hash table and upon successful matching of the hash address from the hash table, the proposed system routes the search request to a service port and provides the …

Bug Management Using Machine Learning, Megha Jindal, Richa Gupta, Subham Mishra, Carlos Arguelles, Babu Prasad Elumalai

Defensive Publications Series

Automated tests of software can often independently log different bugs for the same underlying problem (root cause). Manually identifying duplicate bugs is a source of toil for engineers. A related problem of bug management is that of bug routing, e.g., determining the right team or person to route a bug report to for the purposes of debugging. This disclosure describes techniques for bug deduplication and bug routing based on machine learning (ML). Per the techniques, a binary machine classification model is trained to aggregate bugs with a common root cause. Bugs in a class of bugs with a common root …

Integrating Bug Deduplication In Software Development And Testing, Carlos Arguelles, Megha Jindal, Babu Prasad Elumalai, Subham Mishra, Richa Gupta

Defensive Publications Series

A bug deduplicator identifies independently discovered bugs that have the same underlying cause. Deduplication of bugs reduces toil for the software team by reducing the number of bugs that developers need to examine. However, if a bug deduplicator incorrectly classifies a bug as a duplicate, human developers might ignore the bug, allowing it to escape to production. A tradeoff exists between toil reduction and risk tolerance. This disclosure describes techniques that enable a software team to trade off the effort to remove bugs (e.g., auto-close bugs so that humans save toil and time) against the risk of errors in a …

Speech And Sentiment Analysis To Provide Spoken Language Feedback To Second-Language Learners And Speech Therapy Patients, Chris Pearce, Lasonia Packenham, Tracy Caivano, Estefania Fernandez Alvarez, Sampada Khandekar

Defensive Publications Series

Language instruction and speech therapy are challenging endeavors. The different challenges that are associated with such endeavors have been exacerbated by the impact of coronavirus disease 2019 (COVID-19), where, for example, language instruction and speech therapy are increasingly taking place through teleconferences. To address the types of challenges that were described above, techniques are presented herein that encompass the combination of a phonetic understanding of a speaker's utterance (as understood by, for example, a speech transcription engine coupled with knowledge of a speaker's native language) with a sentiment analysis of a native and/or non-native speaker (obtained by, for example, performing …

Enhanced Host Discovery In Sdn/Fabric-Based Networks, Eric Levy-Abegnoli, Pascal Thubert, Patrick Wetterwald, Ratko Kovacina

Defensive Publications Series

Various solutions are provided herein to facilitate the efficient discovery of hosts in large network environments, such as software-defined networking (SDN) or fabric-based networks, utilizing several techniques. A first technique supports the ability to efficiently manage silent ports and silent media access control (MAC) addresses. This technique involves applying a novel heuristic to ports and MAC addresses, classifying such entities (as silent, quiet, and noisy), and intelligently polling such entities. A second technique supports a Multicast Listener Discovery (MLD)-based host discovery approach that is applicable to Internet Protocol (IP) version 4 (IPv4) and involves a host creating an IP version …

A System And Method For Providing Multiple Transactions For A Single Payment, Samuel Oliver Mr

Defensive Publications Series

Present disclosure provides a system and method for providing multiple transactions by tapping multiple cards consecutively on a Point of Sale (PoS) device (104) for a single payment. The system may include a PoS device (104) and a network server (109). The PoS device (104) receives merchant input for a payment. The merchant (102) input includes total transaction amount and payment split input into the PoS device (104). The payment split input indicates the total number of customers who desire to divide the total transaction amount within themselves. The PoS device (104) splits the total amount by the number input …

Providing Suggestions Of Expanded Text From Abbreviated Text Input, Ajit Narayanan, Meredith Morris, Subhashini Venugopalan, Michael Terry, Sherry Tongshuang Wu, Shanqing Cai, Carrie Cai

Defensive Publications Series

This disclosure describes techniques to provide suggestions of expanded text from abbreviated or compressed text that has been input by a user. A language model is used to determine and present the most likely full words and phrases that match user intent based on the user’s abbreviated text input, such as the first letter of each word of a phrase and/or omission of one or more words of the phrase. The described techniques can greatly improve speed of text entry to devices via a keyboard or other input modality.

Automatically Deriving Spreadsheet Cell Values Using Natural Language, Andy Lavery, Earl J. Wagner, Matthew Albright

Defensive Publications Series

This disclosure describes techniques to populate spreadsheet cells that depend on, but aren’t numerically calculable from, other cells. Based on a natural language query, the empty cells of a partially populated sheet that are contextually dependent on the thus-far-filled cells are automatically filled. The techniques provide greater speed, accuracy, and scalability for datasets small and large, enabling users to efficiently explore data. The techniques obviate the need to manually populate each cell in the sheet, a time-consuming and error prone procedure that does not scale well. Automatically filling in values in a spreadsheet, as described herein, opens up possibilities for …

Automatic Ink Redispersion In The Intermediate Tank By Using Magnetic Fields, Hp Inc

Defensive Publications Series

When a supply is stored, the heavier pigments settle on the bottom part of the supply. Due to that,

prior to installing a new ink supply, it is necessary to do a certain number of shakes to ensure that the

ink is properly mixed.

In some printers, a printhead receives a stream of printing fluid from an Intermediate Tank, which is

supplied with ink from a supply tank by an ink pump. Depending on the Intermediate Tank orientation,

the heavier pigments settle on the deposit.

If the ink is not properly mixed, there would be depleted and enriched ink inside …

Process To Detect Air Leakages In A 3d Printer, Hp Inc

Defensive Publications Series

In 3D printers, the Powder fusion process occurs in a sealed environment called print chamber,

that is kept at negative pressure (relative to client room); the print chamber is sealed not only to

prevent undesired contamination of the client room, but also to avoid unwanted air leaks from

the client room (or by other zone of the 3D printer itself) to the print chamber. In fact, these air

leaks may cause unwanted airflows in the printing area: they cool down only a specific zone of

the printable area, changing the energy balance in this region and therefore affecting Part

Quality. …

A Kubernetes Native Approach To Distribute Webhook Events To Multiple Non-Exported Clusters, Prashanth Kumar, Luis Pabón

Defensive Publications Series

The proposed model is a Kubernetes native approach to distribute webhook events to multiple non-exported clusters. A single Kubernetes service is exposed to listen to external webhooks from a GipOps service. Non-exposed clusters connect to the exposed Kubernetes service to watch for updates from webhook events.