Telefonaktiebolaget lm ericsson (publ) (20240095588). METHODS, APPARATUS AND MACHINE-READABLE MEDIUMS RELATING TO MACHINE LEARNING MODELS simplified abstract

METHODS, APPARATUS AND MACHINE-READABLE MEDIUMS RELATING TO MACHINE LEARNING MODELS

Organization Name

telefonaktiebolaget lm ericsson (publ)

Inventor(s)

Konstantinos Vandikas of Solna (SE)

Aneta Vulgarakis Feljan of Stockholm (SE)

Athanasios Karapantelakis of Solna (SE)

Marin Orlic of Bromma (SE)

Selim Ickin of Stocksund (SE)

METHODS, APPARATUS AND MACHINE-READABLE MEDIUMS RELATING TO MACHINE LEARNING MODELS - A simplified explanation of the abstract

This abstract first appeared for US patent application 20240095588 titled 'METHODS, APPARATUS AND MACHINE-READABLE MEDIUMS RELATING TO MACHINE LEARNING MODELS

Simplified Explanation

The patent application describes a method for determining bias in machine learning models by comparing a local model to a remote model and analyzing biasing data parameters.

• Forming a training dataset with input and output data samples from a remote machine learning model.
• Training a local machine learning model to approximate the remote model.
• Interrogating the local model to identify bias in the remote model with respect to biasing data parameters.

Potential Applications

This technology could be applied in various industries where bias in machine learning models needs to be identified and addressed, such as finance, healthcare, and criminal justice.

Problems Solved

This technology helps in identifying bias in machine learning models, which is crucial for ensuring fairness and accuracy in decision-making processes that rely on these models.

Benefits

The method provides a systematic approach to detect bias in machine learning models, leading to improved transparency and accountability in automated decision-making systems.

Potential Commercial Applications

Companies developing machine learning models for various applications could use this technology to ensure their models are free from bias and comply with regulatory requirements.

Possible Prior Art

Prior art in the field of machine learning may include research papers, patents, or commercial products that address bias detection in machine learning models.

How does this method compare to existing bias detection techniques in machine learning models?

This article does not provide a comparison with existing bias detection techniques, leaving the reader wondering about the advantages and limitations of this specific method.

What are the potential limitations or challenges in implementing this method in real-world scenarios?

The article does not discuss potential challenges or limitations that may arise when applying this method in practical settings, leaving room for uncertainty regarding its feasibility and effectiveness in real-world applications.

Original Abstract Submitted

a method is provided for determining bias of machine learning models. the method includes: forming a training dataset including input data samples provided to a remote machine learning model developed using a machine learning process, and corresponding output data samples obtained from the remote machine learning model; training a local machine learning model which approximates the remote machine learning model using a machine learning process and the training dataset; and interrogating the trained local machine learning model to determine whether the remote machine learning model is biased with respect to one or more biasing data parameters.