17978672. ELECTRONIC DEVICE AND METHOD WITH MACHINE LEARNING TRAINING simplified abstract (SAMSUNG ELECTRONICS CO., LTD.)

ELECTRONIC DEVICE AND METHOD WITH MACHINE LEARNING TRAINING

Organization Name

SAMSUNG ELECTRONICS CO., LTD.

Inventor(s)

Minsu Ko of Suwon-si (KR)

Sungjoo Suh of Seongnam-si (KR)

Huijin Lee of Pohang-si (KR)

Eunju Cha of Suwon-si (KR)

ELECTRONIC DEVICE AND METHOD WITH MACHINE LEARNING TRAINING - A simplified explanation of the abstract

This abstract first appeared for US patent application 17978672 titled 'ELECTRONIC DEVICE AND METHOD WITH MACHINE LEARNING TRAINING

Simplified Explanation

The abstract describes a method for training a machine learning model on an electronic device. The device includes a memory storing a discriminator model, and a processor configured to perform various operations on training images.

• The processor extracts two overlapping image patches from a training image.
• It then extracts feature maps from these patches based on a layer of the discriminator model.
• A partial feature map is extracted from a projected map, which is generated based on the first feature map.
• The discriminator model is trained using an objective function value generated from the partial feature maps.

Potential Applications:

• This technology can be applied in various fields where machine learning models are used, such as computer vision, natural language processing, and speech recognition.
• It can be used for tasks like image classification, object detection, and sentiment analysis.

Problems Solved:

• The method provides a way to train a machine learning model by extracting and utilizing feature maps from overlapping image patches.
• It improves the training process by incorporating partial feature maps and projected maps.

Benefits:

• The use of overlapping image patches and feature maps can enhance the model's ability to capture fine-grained details and patterns.
• Training the model with partial feature maps and projected maps can improve its performance and accuracy.
• The method allows for efficient training of the model on electronic devices with limited memory and processing capabilities.

Original Abstract Submitted

Training of a machine learning model is included. An electronic device includes a memory storing a machine learning model including a discriminator model, and a processor configured to extract, from a training image, a first image patch and a second image patch at least partially overlapping the first image patch, extract, from the first image patch, a first feature map based on a layer of the discriminator model, extract, from the second image patch, a second feature map based on the layer, extract a first partial feature map from a projected map that is projected based on the first feature map, and train the discriminator model based on a first objective function value generated based on a second partial feature map and the first partial feature map, wherein the second partial feature map corresponds to a part of the extracted second feature map.