International Business Machines Corporation

Horst Cornelius Samulowitz

Gregory Bramble

Dakuo Wang

Saket Sathe

Kavitha Srinivas

Deepak Srinivas Turaga

Peter Daniel Kirchner

Chandrasekhara K Reddy

Udayan Khurana

Ashish Sabharwal

Charu Chandra Aggarwal

Jing James Xu

Bei Chen

Chuang Gan

Si Er Han

Ji Hui Yang

Gerald James Tesauro

Tamir Klinger

Long H Vu

Dhavalkumar C Patel

Syed Yousaf Shah

Xuan-Hong Dang

Parikshit Ram

Arunima Chaudhary

Petr Novotny

Julian Timothy Dolby

Jun Song Wang

Bo Zhang

Petros Zerfos

Oktie Hassanzadeh

Lisa Amini

Sijia Liu

Jun Wang

Wesley M Gifford

Vijay Anand Saraswat

Deepak Vijaykeerthy

Vaibhav Saxena

Abel Nicolas Valente

William Karol Lynch

Shaikh Shahriar Quader

Alexander Gray

Daniel Karl I Weidele

Elias Khalil

Fabio Lorenzi

Long Vu

Andrea Loreggia

Yuri Malitsky

Sainyam Galhotra

Fatemeh Nargesian

Chad Marston

Aindrila Basak

Gregory Filla

Carolina Maria Spina

A dataset including features and values associated with the features can be received. Each of the features in the dataset can be mapped to a corresponding node in a knowledge graph based on the concept represented by the corresponding node. The knowledge graph can be traversed to find a candidate node connected to at least one mapped node, the candidate node not being mapped to a feature in the dataset. A concept associated with the candidate node can be identified as a new feature. A machine learning model pipeline can use the features in the dataset and the new feature to select a subset of features for training a machine learning model.

Interactive feature engineering in automatic machine learning with domain knowledge

The exemplary embodiments disclose a method, a computer program product, and a computer system for determining that one or more model pipelines satisfy one or more constraints. The exemplary embodiments may include detecting a user uploading data, one or more constraints, and one or more model pipelines, collecting the data, the one or more constraints, and the one or more model pipelines, and determining that one or more of the model pipelines satisfies all of the one or more constraints based on applying one or more algorithms to the collected data, constraints, and model pipelines.

Machine learning with multiple constraints

Techniques for computer software code analysis are disclosed. One or more data flows are generated, based on analyzing software code using static analysis. A data object is identified in the software code using the one or more data flows, the data object relating to a structured dataset. A correspondence between a code expression in the software code and a characteristic of the structured dataset is identified, based on analyzing one or more reads from and one or more writes to the data object using the one or more data flows. The code expression for the structured dataset is analyzed, based on the correspondence, including at least one of: (i) generating a software code recommendation engine based on the code expression and the structured dataset, or (ii) generating one or more lambda expressions for application to the structured dataset, based on the code expression.

Mining code expressions for data analysis

To automate time series forecasting machine learning pipeline generation, a data allocation size of time series data may be determined based on one or more characteristics of a time series data set. The time series data may be allocated for use by candidate machine learning pipelines based on the data allocation size. Features for the time series data may be determined and cached by the candidate machine learning pipelines. Predictions of each of the candidate machine learning pipelines using at least the one or more features may be evaluated. A ranked list of machine learning pipelines may be automatically generated from the candidate machine learning pipelines for time series forecasting based upon evaluating predictions of each of the one or more candidate machine learning pipelines.

Automated time series forecasting pipeline generation

A processor may receive structured data. The structured data may include one or more columns and associated column names. The processor may analyze the structured data. Analyzing the structured data may include gathering a requisite set of keywords from the associated column names across all columns and/or a sample of column cells. The processor may access a corpus of documents. Each of the documents in the corpus may be associated with a respective keyword. The processor may search the corpus of documents based on the requisite set of keywords. The processor may summarize one or more documents associated with the requisite set of keywords.

Automatic domain annotation of structured data

Methods, computer program products and/or systems are provided that perform the following operations: obtaining a performance matrix representing accuracies obtained by executing a plurality of pipelines on a plurality of training data sets, wherein a pipeline comprises a series of operations performed on a data set; selecting a defined number of top pipelines as potential pipelines for a testing data set based, at least in part, on a similarity between the testing data set and each of the plurality of training data sets represented in the performance matrix; storing results from executing each of the potential pipelines as a new data set; determining a pipeline accuracy for each of the potential pipelines when executed against the testing data set; and providing a recommended pipeline for use with the testing data set based, at least in part, on the pipeline accuracy for each potential pipeline.

Automated machine learning using nearest neighbor recommender systems

Computer hardware and/or software that performs the following operations: (i) assessing a performance of a plurality of unsupervised machine learning pipelines against a plurality of data sets; (ii) associating the performance with meta-features corresponding to respective pipeline/data set combinations; (iii) training a supervised meta-learning model using the associated performance and meta-features as training data; and (iv) utilizing the trained model to identify one or more pipelines for processing an input data set.

Automated unsupervised machine learning utilizing meta-learning

An embodiment of the invention may include a method, computer program product, and system for creating a data analysis tool. The method may include a computing device that generates an AI pipeline based on an input dataset, wherein the AI pipeline is generated using an Automated Machine Learning program. The method may include converting the AI pipeline to a non-native format of the Automated Machine Learning program. This may enable the AI pipeline to be used outside of the Automated Machine Learning program, thereby increasing the usefulness of the created program by not tying it to the Automated Machine Learning program. Additionally, this may increase the efficiency of running the AI pipeline by eliminating unnecessary computations performed by the Automated Machine Learning program.

Code generation for Auto-AI

A quality determination method, system, and computer program product that includes performing a dimensionality reduction on a high-dimensional dataset to form a dimensional-reduced dataset and determining, using a machine learning tool executed on a computing device, a quality of the dimensional-reduced dataset via a review of an extracted feature extracted from the dimensional-reduced dataset.

Quality assessment of extracted features from high-dimensional machine learning datasets

A system that provides a mathematical formulation for new problem of model validation and model selection in presence of test data feedback. The system comprises a memory that stores computer-executable components. A processor, operably coupled to the memory, executes the computer-executable components stored in the memory. A selection component selects a metric of performance evaluation accuracy; and a configuration component configures performance evaluation schemes for machine learning algorithms. A characterization component employs a supervised learning-based approach to characterize relationship between the configuration of the performance evaluation scheme and fidelity of performance estimates; and an optimization component that optimizes accuracy of the machine learning algorithms as a function of size of training data set relative to size of validation data set through selection of values associated with the configuration parameters.

Methods for automatically configuring performance evaluation schemes for machine learning algorithms

A method, system, and computer program product are disclosed. The method includes extracting at least one identifier from a formula in a document and extracting text passages in the document that contain the identifier(s). The method also includes selecting an identifier and extracted text passages containing the identifier, as well as generating identifier-passage pairs for the selected text passages and the identifier. Further, the method includes submitting the identifier-passage pairs to a question answering (QA) model, which generates candidate answers from the selected text passages. A definition of the identifier is then selected from the candidate answers.

Question answering approach to semantic parsing of mathematical formulas

Techniques regarding one or more automated machine learning processes that analyze time series data are provided. For example, one or more embodiments described herein can comprise a system, which can comprise a memory that can store computer executable components. The system can also comprise a processor, operably coupled to the memory, and that can execute the computer executable components stored in the memory. The computer executable components can comprise a time series analysis component that selects a machine learning pipeline for meta transfer learning on time series data by sequentially allocating subsets of training data from the time series data amongst a plurality of machine learning pipeline candidates.

Automated machine learning pipeline generation

Embodiments relate to a system, program product, and method for employing feature engineering to improve classifier performance. A first machine learning (ML) model with a first learning program is selected. The first selected ML model is operatively associated with a first structured dataset. First features in the first dataset directed at performance of the selected ML model are identified. A second structured dataset is assessed with respect to the identified features in the first dataset, and new features in the second dataset are identified, where the new features are semantically related to the identified features in the first dataset. The first dataset is dynamically augmented with the identified new features in the second dataset. The dynamically augmented first dataset is applied to the selected ML model to subject an embedded learning algorithm of the selected ML model to training using the augmented first dataset.

Knowledge aided feature engineering

Embodiments for automated feature engineering by one or more processors are described. One or more selected transformations may be applied to a set of features in a dataset to create a set of transform features using random feature transformation forest (RFTF) classifiers. A transform feature may be selected from the set of transform features having a highest discriminative power as compared to other features of the set of transform features. At each node in a decision tree, store the selected feature, a split value, and the one or more selected transformations for the transform feature.

Random feature transformation forests for automatic feature engineering

Embodiments for feature engineering by one or more processors are described. A plurality of transformations are applied to a set of features in each of a plurality of datasets. An output of each of the plurality of transformations is a score. For each of the sets of features, selecting those of the plurality of transformations for which said score is above a predetermined threshold. A signal representative of said selection is generated.

Methods and systems for feature engineering

Growing community of inventors

White Plains, NY, United States of America

Horst Cornelius Samulowitz