International Business Machines Corporation

Hewlett Packard Enterprise Development LP

Kalapriya Kannan

Sameep Mehta

George Thayyil Jacob Sushil

Vijay Mann

Shivkumar Kalyanaraman

Suparna Bhattacharya

Anuradha Bhamidipaty

Madhu K Chetuparambil

Douglas L Voigt

Anil Kumar Vishnoi

Anilkumar Vishnoi

Manish Satyapal Gupta

Biplav Srivastava

Kush Raj Varshney

Pranay Kumar Lohia

Rema Ananthanarayanan

Manish Sethi

Soudip Roy Chowdhury

Srikanta Bedathur Jagannath

Sagar Venkappa Nyamagouda

Jayasankar Nallasamy

Chinmay Chaturvedi

Chirag Talreja

Prasad Pimplaskar

Gandhi Sivakumar

Srikanth Govindaraj Tamilselvam

Ramya Raghavendra

Soma Shekar Naganna

Ravi Kothari

Bruce Rabe

Berthold Reinwald

Vikas Agarwal

Kerrie Lamont Holley

Vijay Arya

John Thomas Richards

Albee Jhoney

Aleksandra Mojsilovic

Darrell Christopher Reimer

Ram Viswanathan

Rachel K E Bellamy

Sumit Mittal

Manish Kesarwani

Manjit Singh Sodhi

Senthil Kumar Kumarasamy Mani

Matthew R Arnold

Hariharan Krishna

Kaoutar El Maghraoui

Neeraj Ramkrishna Singh

Girish Bhimrao Chafle

Michael James Hind

Hima Patel

Samiulla Zakir Hussain Shaikh

Stephanie Houde

Narendran Sachindran

David John Piorkowski

Shettigar Parkala Srinivas

Balaji Ganesan

Jason Tsay

Sai Koti Reddy Danda

Muthukumar Murugan

Samuel Chung Hoffman

Murali Nidugala

Poornima Iyengar

Krishnaraju Thangaraju

Karan Rai

George T Jacob

Chaitra Kallianpur

Sumanth Tummala

One embodiment provides a computer implemented method, including: obtaining an information document corresponding to an entity, wherein the information document includes redacted information spans; identifying an entity type for each of the redacted information spans, wherein the entity type identifies a relationship between a redacted information span and at least one other entity within the information document; replacing the redacted information spans with replacement entities corresponding to the entity type of a given redacted information span, wherein the replacing is performed in view of a frequency distribution of actual information and wherein the replacing includes maintaining relationships of the redacted information spans; and controlling bias within the replacement entities, wherein the controlling includes detecting bias within the replacement entities.

Generation of training data from redacted information

Systems and methods are provided for generating extract-transform-load ('ETL') machine learning (“ML”) pipeline validation rules based on user-input, wherein the ETL ML pipeline validation rules may be applicable to validate an ETL ML pipeline against multiple test datasets. The ETL ML pipeline validation rules may comprise compute-type validation rules for computing expected values of data structures within a dataset output by the ETL ML pipeline. The ETL ML pipeline validation rules may comprise check-type validation rules for checking whether data structures within a dataset output by the ETL ML pipeline have intended characteristics.

Data correctness and validation using validation definition language

Examples described herein relate to preparing datasets in a storage device for machine learning (ML) applications. Examples include maintaining ML facet mappings between ML facets and dataset preparation tags, deriving ML facets of a dataset stored in the storage device, and generating filtered datasets from the datasets using the ML facets and ML facet mappings. The filtered dataset is associated with improved dataset quality compared to unfiltered dataset. The storage device transmits the filtered dataset to ML applications requesting the dataset. Some examples include recommending, by the storage device, ML facets to the ML application based on performance metrics.

Machine learning facets for dataset preparation in storage devices

Systems and methods are provided for reusing machine learning models. For example, the applicability of prior models may be compared using one or more assessment values, including a similarity threshold and/or an accuracy threshold. The similarity threshold may identify a similarity of data between a first data set used to generate a first model and a new data set that is received by the system. When the similarity between these two data sets is exceeded, the system may reuse a model with the highest similarity value. When an accuracy value of the data set does not exceed an accuracy threshold, the system may initiate a retraining process to generate a second ML model associated with the second data.

Artificial intelligence optimization platform

Methods, systems, and computer program products for factchecking artificial intelligence models using blockchain are provided herein. A computer-implemented method includes obtaining at least one artificial intelligence model and at least one set of data related to the at least one artificial intelligence model; determining a set of characteristics based at least in part on the at least one artificial intelligence model and the at least one set of data; selecting one of a plurality of networks based at least in part on a target deployment of the at least one artificial intelligence model to verify the set of characteristics; generating a report based at least in part on verifying the set of characteristics using the selected network, wherein the report establishes a threshold level of trust for the at least one artificial intelligence model; and storing the report on a blockchain.

Factchecking artificial intelligence models using blockchain

Examples include bypassing a portion of an analytics workflow. In some examples, execution of an analytics workflow may be monitored upon receipt of a raw data and the execution may be interrupted at an optimal bypass stage to obtain insights data from the raw data. A similarity analysis may be performed to compare the insights data to a stored insights data in an insights data repository. Based, at least in part, on a determination of similarity, a bypass operation may be performed to bypass a remainder of the analytics workflow.

Similarity analyses in analytics workflows

Systems and methods are provided for generating extract-transform-load ('ETL') machine learning (“ML”) pipeline validation rules based on user-input, wherein the ETL ML pipeline validation rules may be applicable to validate an ETL ML pipeline against multiple test datasets. The ETL ML pipeline validation rules may comprise compute-type validation rules for computing expected values of data structures within a dataset output by the ETL ML pipeline. The ETL ML pipeline validation rules may comprise check-type validation rules for checking whether data structures within a dataset output by the ETL ML pipeline have intended characteristics. Where the ETL ML pipeline validation rules are applicable to validate an ETL ML pipeline against a test dataset which was not referenced to describe the ETL ML pipeline validation rules, then the ETL ML pipeline may reuse these ETL ML pipeline validation rules to validate the ETL ML pipeline without further user-input.

Automatic expected validation definition generation for data correctness in AI/ML pipelines

A method for blockchain certification of artificial intelligence factsheets that includes receiving by a computing device, an artificial intelligence model. The computing device generates an artificial intelligence factsheet based upon logic of the artificial intelligence model. The computing device generates a blockchain link for a blockchain. The blockchain link certifies the artificial intelligence factsheet. The computing device transmits the blockchain link certifying the artificial intelligence factsheet to other computing devices.

Artificial intelligence certification of factsheets using blockchain

Ranking a group of datasets using a computer includes determining a set of target data fields from a set of process documents that indicate user data field preferences. A set of target dataset attributes from a set of data use documents indicate user data scope preferences. A plurality of metadata sets for an associated plurality of datasets the computer determines having a field suitability value exceeding a predetermined suitability threshold value. The FSV represents a degree of similarity between a set of fields associated with said dataset and the set of target data fields. The computer assesses metadata sets with regard to the target attributes and generates a compared attribute score for each candidate dataset. A degree of likelihood is indicated that an associated dataset will have content exhibiting said target dataset attributes. The computer candidate datasets is based on the compared attribute score.

Ranking datasets based on data attributes

Embodiments for determining processing performed on a data element are provided. A file system call associated with a data element stored in a storage system is detected. The file system call is analyzed. Data lineage for the data element is determined based on the analyzing of the file system call.

Methods and systems for data traceability and provenance

A method for automatically generating documentation for an artificial intelligence model includes receiving, by a computing device, an artificial intelligence model. The computing device accesses a model facts policy that indicates data to be collected for artificial intelligence models. The computing device collects artificial intelligence model facts regarding the artificial intelligence model according to the model facts policy. The computing device accesses a factsheet template. The factsheet template provides a schema for an artificial intelligence model factsheet for the artificial intelligence model. The computing device populates the artificial intelligence model factsheet using the factsheet template with the artificial intelligence model facts related to the artificial intelligence model.

Generation and management of an artificial intelligence (AI) model documentation throughout its life cycle

Methods, systems, and computer program products for generating a framework for prioritizing machine learning model offerings via a platform are provided herein. A computer-implemented method includes processing, via a computing platform, a machine learning model input by a first user and metadata corresponding to the machine learning model input by the first user; automatically comparing, via the computing platform, the metadata corresponding to the machine learning model with metadata corresponding to one or more existing machine learning models stored by the computing platform; automatically calculating, via the computing platform, initial pricing information for the machine learning model based on the comparison; and outputting, via an interactive user interface of the computing platform, the machine learning model to one or more additional users for purchase in accordance with the calculated initial pricing information.

Generating a framework for prioritizing machine learning model offerings via a platform

One embodiment provides a method, including: receiving, from a user, (i) a dataset and (ii) an intended output from the dataset that is generated in view of a given analytical framework for the dataset, wherein the intended output identifies an output that the user wants from the dataset and wherein the dataset is related to an analytical domain; identifying a plurality of dataset functions related to the analytical domain; determining one or more dataset functions for each of one or more operations identified, wherein the one or more operations are identified using the repository to identify operations used to result in an intended output similar to the received intended output; and recommending an ordered subset of the one or more dataset functions to be used to transform the dataset to the intended output, wherein the ordered subset comprises (i) one dataset function for each of the one or more operations and (ii) an order for performing the one or more operations.

Operations to transform dataset to intent

Some examples relate to associating an insight with data. In an example, data may be received. A determination may be made that data type of the data is same as compared to an earlier data. An insight generated from the earlier data may be identified, wherein the insight may represent intermediate or resultant data generated upon processing of the earlier data by an analytics function, and wherein during generation metadata is associated with the insight. An analytics function used for generating the insight may be identified.

Associating insights with data

Growing community of inventors

Bengaluru, India

Kalapriya Kannan