17436927. METHOD AND APPARATUS FOR SEMI-SUPERVISED LEARNING simplified abstract (SAMSUNG ELECTRONICS CO., LTD.)

METHOD AND APPARATUS FOR SEMI-SUPERVISED LEARNING

Organization Name

SAMSUNG ELECTRONICS CO., LTD.

Inventor(s)

Ivana Balazevic of Staines (GB)

Carl Allen of Staines (GB)

Timothy Hospedales of Staines (GB)

METHOD AND APPARATUS FOR SEMI-SUPERVISED LEARNING - A simplified explanation of the abstract

This abstract first appeared for US patent application 17436927 titled 'METHOD AND APPARATUS FOR SEMI-SUPERVISED LEARNING

Simplified Explanation

The patent application describes a computer-based method for training a machine learning model using both labelled and unlabelled data. The method involves two main steps: training a loss module and training a task module.

• The method starts by obtaining a set of training data that includes both labelled data items (data items with known labels) and unlabelled data items (data items without labels).
• The loss module of the machine learning model is trained using the labels in the set of labelled data items. This trained loss module is capable of estimating the likelihood of a label for a given data item.
• The task module of the machine learning model is then trained using the trained loss module, the set of labelled data items, and the set of unlabelled data items. This trained task module is capable of making predictions of labels for input data.

Potential applications of this technology:

• Improving the accuracy and performance of machine learning models by utilizing both labelled and unlabelled data.
• Enhancing the capabilities of predictive models in various domains such as image recognition, natural language processing, and recommendation systems.

Problems solved by this technology:

• Addressing the challenge of limited availability of labelled data, which is often expensive and time-consuming to obtain.
• Overcoming the limitations of traditional supervised learning methods that heavily rely on labelled data.

Benefits of this technology:

• Enables the utilization of unlabelled data to enhance the training process and improve the performance of machine learning models.
• Reduces the dependency on large amounts of labelled data, making it more cost-effective and efficient to train models.
• Enhances the accuracy and reliability of predictions made by machine learning models.

Original Abstract Submitted

Provided is a computer-implemented method for training a machine learning (ML) model using labelled and unlabelled data, the method comprising obtaining a set or training data comprising a set of labelled data items and a set of unlabelled data items, training a loss module of the ML model using labels in the set of labelled data items, to generate a trained loss module capable of estimating a likelihood of a label for a data item, and training a task module of the ML model using the loss module, the set of labelled data items, and the set of unlabelled data items, to generate a trained task module capable of making a prediction of a label for input data.