International Business Machines Corporation

Kyndryl, Inc.

Angelo Danducci, Ii

Igor S Ramos

Kimberly J Taft

Devon E Mensching

Denny Nguyen

Hilary Ricky

Andrew S Cassidy

Marc Dickenson

Jean Xu Yu

Stefan A G Van Der Stockt

Anindita Das

Kirti A Apte

Keri Wheatley

Harshan Kumar Yennamaneni

Jean Elvira Schunck

Brittany L Barnes

Tyrome Sweet

Olga Saprycheva

A computer-implemented method, computer program product, and computer system to classify executing processes as malicious based on opcode data. A processor to retrieves live opcode data corresponding to one or more processors executing a process. A processor generates a malicious classification of the process based on output from a machine learning model that is provided the live opcode data. A processor, in response to the machine learning model classifying the process as malicious, restricts the execution of the process on the one or more processors.

Detection of malicious encryption based on machine learning

A processor may generate a workstation characteristics model of a workspace. A processor may generate respective user profiles for each of one or more users. Each of the respective user profiles may include user data associated with a respective user. A processor may assign, based on the respective user profiles, a respective risk level to each of the respective user profiles. A processor may compare each of the respective risk levels to a risk threshold level. A processor may apply a user characteristic model. A user characteristic model may be based on each of the respective risk levels and the workstation characteristics model. A processor may assign each of the one or more users to respective workstations based on the user characteristic model.

Workstation assignment

A mechanism is provided in a data processing system for role-based cross data source actionable conversation summarization. The mechanism aggregates conversation data from a plurality of conversation data sources. The mechanism applies a computerized summarization process to the aggregated conversation data to generate at least one role-based summary of the aggregated conversation data. The mechanism applies a machine learning classifier to the at least one role-based summary to determine if each sentence in the at least one role-based summary is an actionable item. Responsive to detecting an actionable item, the mechanism adds the actionable item to the at least one role-based summary.

Role-based cross data source actionable conversation summarizer

According to one embodiment, a method, computer system, and computer program product for personalizing playback of an audio stream is provided. The present invention may include sectioning the audio stream into one or more content blocks and one or more filler blocks; determining one or more topics associated with the one or more filler blocks; determining a level of complexity associated with the one or more filler blocks; determining a listener's level of interest in and level of comprehension of the one or more topics; based on the level of complexity, the level of interest, and level of comprehension corresponding to the one or more topics, assigning a playback speed to the one or more content blocks associated with the one or more topics; and modifying the one or more content blocks to play at the assigned playback speed.

Individualized media playback pacing to improve the listener's desired outcomes

Provided is a method for relocating a solar power unit in response to a redeployment event. A first location of a deployed solar power unit may be determined. A processor may detect a redeployment event for the solar power unit at the first location. In response to the redeployment event, the processor may determine a new location for the solar power unit. The method may further comprise relocating the solar power unit to the new location.

Autonomous solar power system

Provided is a method for deploying multiple solar power units to a location. The method comprises determining a location to which a solar power unit is deploying. The method further comprises determining that a second solar power unit is deploying to a same location as the solar power unit. A redeployment plan for the second solar power unit is generated. The redeployment plan is provided to the second solar power unit. The method further comprising deploying the solar power unit to the location.

A temperature state of one or more fluids in one or more enclosures is determined. Acoustic data is collected using a microphone. Frequency domain features are determined based on the acoustic data, where the frequency domain features are obtained across a frequency range of the acoustic data. The frequency domain features are correlated to the output of a machine learning acoustic analysis model and the one or more fluids are classified based on temperature based on the correlated one or more frequency domain features. In addition, the machine learning acoustic analysis model is trained by recording acoustic training data using a microphone and collecting temperature data for a fluid across a range of temperatures for the fluid. Frequency domain features are determined across a frequency range of the acoustic training data and the frequency domain features are correlated to the temperature data for a fluid.

Acoustic fluids monitoring

In an approach to estimating product inventory count, one or more computer processors receive one or more images of one or more products residing on a product storage location from an image capturing device. Based on the received images, one or more computer processors determine a count of the one or more products. One or more computer processors determine a confidence in the count of the one or more products. In response to determining the confidence is below a threshold, one or more computer processors calculate a recommended position of the image capturing device to produce an improved image of the one or more products. One or more computer processors transmit instructions to the image capturing device to move to the recommended position. One or more computer processors determine whether the image capturing device is in the recommended position.

Image-based inventory estimation

In an approach to anonymous inventory tracking, one or more computer processors receive a first image of one or more products from a user. The one or more computer processors remove data associated with an identity of the user from the first image. The one or more computer processors determine that a storage location associated with the one or more products in the first image is detected. The one or more computer processors determine an inventory count of the one or more products in the first image. The one or more computer processors submit the first image and the inventory count of the one or more products in the first image to a blockchain ledger in association with the storage location. The one or more computer processors take an action based on the inventory count.

Anonymous inventory tracking system

In an approach for recommending locations to visit at an optimal experience time and optimal spot to take a photo and/or a video of a location, a processor determines a geographic location where a user is located. A processor identifies a plurality of targeted locations within a pre-defined distance of the geographic location of the user. A processor determines an optimal time to visit the plurality of targeted locations. A processor determines an optimal weather condition to experience the plurality of targeted locations. A processor determines an optimal spot to capture at least one of a photo and a video of the plurality of targeted locations. A processor assigns a ranking to the plurality of targeted locations based on a set of factors in a user profile of the user. A processor outputs a recommendation with at least one of the plurality of targeted locations as an alert notification.

Recommending targeted locations and optimal experience time

An embodiment includes calculating an efficacy value of a context assigned to an asset, the efficacy value being based at least in part on a percentage of assets assigned to the context. The embodiment also includes calculating, responsive to the calculated efficacy value being within a predetermined range, a normalized relevance value of a first tag associated with the asset. The embodiment also includes generating relative relevance value for the first tag and a reason code associated with a basis for the relative relevance value of the first tag based at least in part on the normalized relevance value of the first tag. The embodiment also includes initiating, responsive to generating the relative relevance value and the reason code for the first tag, a database command on a training dataset stored in the database that includes the first tag.

Ranking of asset tags

An approach is provided for using a machine learning model to reduce food waste. Estimation models for food items are created by using a machine learning model. The estimation models have prediction functions specifying timelines during which the food items are not spoiled and are in a condition for consumption. Prediction function solutions are generated for a food item inventory to indicate menu items for respective time periods. Each menu item includes food item(s) which are in condition for consumption based on the estimation models. Menu recommendations corresponding to the time periods are generated. Each menu includes menu item(s) indicated by the prediction function solutions. The menu recommendations reduce a waste of the food items through spoilage by increasing a likelihood of consumption of the food items within the respective timelines.

Reducing food waste by using a machine learning model

A method utilizes Artificial Intelligence (AI) system to broadcast a real-time event to a particular broadcast audience. The AI system receives a profile of a particular broadcast audience for a real-time event, and concurrent real-time Audio/Video (/V) streams for a particular portion of the real-time event. The AI system ranks the concurrent real-time A/V streams for the particular portion of the real-time event based on the types of event activities depicted in each of the concurrent real-time A/V streams and the profile of the particular broadcast audience for the real-time event; selects a particular real-time A/V stream, from the concurrent real-time A/V streams, for the particular broadcast audience based on the ranking; and directs a performance of a broadcast action related to the selected particular real-time A/V stream.

Growing community of inventors

Austin, TX, United States of America

Angelo Danducci, Ii