International Business Machines Corporation

Nuance Communications, Inc.

Girish Dhanakshirur

Ajay Chebbi

Chethan Kumar N Settipalli

Rocco F Palermo, Jr

Kuntal Dey

Baiju Dhirajlal Mandalia

James Robert Lewis

Sougata Mukherjea

Sumit Mittal

Anton Aleksandrov

Alwyn R Lobo

Jeffrey S Kobal

Jayakarthik Jayabalan

Bharat Devdas

Srihari Kulkarni

Venkatakrishnan Sampath Kumaran

Thejaswini Ramachandra

Larissa Guerrero

Daniel Yellin

Wendi L Nusbickel

Saraswathi Sailaja Perumalla

Hemant Kumar Sivaswamy

Shinoj Zacharias

Leslie Robert Wilson

Josephine E Justin

Anoop G M Ramachandra

Srinivasan Nanduri

Rachana Vishwanathula

Vidya Chandrashekar

Lenora E Wright

Aimee Silva

Deepak Rai

Norton Samuel A Stanley

Jason Garrabrant

Andrew Otte

Melissa Schoonmaker

A method, computer program product, and computer system for network topology optimization. A graph convolutional network (GCN) model is trained using training network topology datasets as input. Each training network topology dataset includes: (i) a specified network topology and an associated optimal network topology and an optimality function value for the optimal network topology, (ii) relative weights of optimality parameters including performance, availability, and scalability, and (iii) an identification of an optimality function of the optimality parameters weighted by the relative weights. The specified network topology includes components, relationships between the components, and metadata pertaining to the components. Each output node in an output layer of the GCN model includes an optimality function value for a different candidate network topology. One of the output nodes identifies an optimum network topology relative to the specified network topology as being the candidate network topology having a highest optimality function value.

Cloud topology optimization using a graph convolutional network model

Distributing machine learning models to electronic devices is provided. A number of machine learning models are built and stored on a server, wherein each model is customized to different device and user requirements. A request for a machine learning model is received from an electronic device, wherein the request includes a device profile comprising device attributes and user characteristics. The server determines which machine learning model from among the number of models is best suited for the device profile according to a set of model policies. The machine learning model best suited for the device profile is selected and downloaded to the device.

Continuously improve recognition or prediction accuracy using a machine learning model to train and manage an edge application

Provided are techniques for synchronizing a device using push notifications. For each of a plurality of messages, a message record is created in a table of a database with a message create time for a message, a device identifier of a device to receive the message, an application identifier of an application on the device to receive the message, message content of the message, and an indicator to indicate whether the message has been picked up by the device. A message is selected from the plurality of messages. It is determined that a message retry count for the message has not been met and the message has not been picked up by the device. It is determined that it is time to retry sending a silent push notification based on a retry interval. The silent push notification is sent to the application on the device to wake up the application.

Synchronizing a device using push notifications

Gathering and verifying customer information via SMS and in-person representatives. A hash is received with customer account information from a business correspondent application, wherein the account information is given in-person to a business correspondent, the business correspondent enters the account information into the business correspondent app, the account information is verified by the business correspondent after entering the account information in the business correspondent app, and the hash is generated by the business correspondent app after the business correspondent app receives confirmation of verification of the account information from the business correspondent. An account number is generated for the customer. The account number is sent to the business correspondent app and to the customer. An enrollment PIN is sent to the customer, wherein the enrollment PIN is linked to the account number. The business correspondent app is synced after the business correspondent app receives the enrollment PIN from the customer.

Mobile based multi-channel citizen account origination in digital economy

Gathering and verifying customer information via SMS and in-person representatives. Customer account information is received, wherein the customer account information is given to a business correspondent by a customer. A verification of the customer account information is received, wherein the verification is performed by the business correspondent. A hash is generated with the customer account information. The hash is sent to a data center. An account number is received from the data center. An enrollment PIN is received, wherein the enrollment PIN is sent from the data center to the customer, the customer gives the enrollment PIN to the business correspondent, and the business correspondent enters the enrollment PIN into the business correspondent app, and wherein the enrollment PIN is linked to the account number. The data center is synced after receiving the enrollment PIN.

An approach is provided for tuning middleware. Performance-related settings are loaded. Performance data of the middleware of a MobileFirst Platform (MFP) running in a docker container infrastructure is received. The performance data is collected by agents installed in container groups. Based on the performance data, a performance issue in one of the container groups is identified and a server included in the one container group is identified as a source of the performance issue. Recommendations are generated for tuning the middleware by modifying one or more of the performance-related settings. While the middleware is running in the docker container infrastructure, one of the recommendations is applied to modify one of the performance-related settings which dynamically tunes the middleware, thereby resolving the performance issue.

Automatically tuning middleware in a mobilefirst platform running in a docker container infrastructure

In an approach for managing updates of a mobile software application, a computer receives information regarding a mobile computing device and a mobile software application on the mobile computing device, wherein the information regarding the mobile computing device comprises at least one or more of a location of the mobile computing device, a network connection type, a security bit associated with the mobile computing device, and a checksum associated with the at least one mobile software application. The computer determines whether the mobile software application matches a stored mobile software application on a computing device. Responsive to determining that the mobile software application does not match the stored mobile software application on the computing device, the computer determines an update policy for scheduling an update to the mobile software application on the mobile computing device.

Intelligent mobile application update

A method for gathering and verifying customer information via SMS and in-person representatives are provided. The method may include a business correspondent receiving and verifying customer information in-person; sending the customer information to a data center, via a business correspondent application (BC app), to create a customer account; sending the account information to the customer and the BC app; verifying the account information with the customer; and syncing the BC app with the data center.

A method, system, and/or computer program product downloads a blob of data onto a mobile device. A blob of data that is to be downloaded from a server to a mobile device via a particular network is identified. A first maximum chunk size for the first network for the particular network is identified, as is a timeout value for the mobile device when downloading a chunk of data. An optimal chunk size for chunking the blob of data is determined based on the timeout value of the mobile device and the maximum chunk size for the first network. The blob of data is partitioned into chunks of data based on the first optimal chunk size, and the chunks of data are transmitted from the server to the mobile device.

Network-specific data downloading to a mobile device

A contextual push notification service is provided, including: receiving, by a mobile device, a notification message. The notification message comprises at least one context rule and a notification message. The at least one context rule is compared, on the mobile device, with at least one user context preference. Based on the results of the comparison, a notification action is taken.

Contextual push notification service

A method for execution on a server for serving presence information, the method for providing dynamically loaded speech recognition parameters to a speech recognition engine, can be provided. The method can include storing at least one rule for selecting speech recognition parameters, wherein a rule comprises an if-portion including criteria and a then-portion specifying speech recognition parameters that must be used when the criteria is met. The method can further include receiving notice that a speech recognition session has been initiated between a user and the speech recognition engine. The method can further include selecting a first set of speech recognition parameters responsive to executing the at least one rule and providing to the speech recognition engine the first set of speech recognition parameters for performing speech recognition of the user.

Automated tuning of speech recognition parameters

A method of analyzing speech application performance can include determining a call path for each of a plurality of calls from a log of an interactive voice response system having a speech application. Each call path can be defined by an ordered set of dialog nodes of the speech application. A number of occurrences of at least one selected type of event for the dialog nodes of the plurality of calls can be counted. At least one call path from the plurality of call paths can be indicated according, at least in part, to the number of occurrences of the selected type of event within the at least one indicated call path.

Analyzing speech application performance

Growing community of inventors

Bengaluru, India

Girish Dhanakshirur