18103887. APPARATUS AND METHOD WITH NEURAL NETWORK IMPLEMENTATION OF DOMAIN ADAPTATION simplified abstract (SAMSUNG ELECTRONICS CO., LTD.)
Contents
APPARATUS AND METHOD WITH NEURAL NETWORK IMPLEMENTATION OF DOMAIN ADAPTATION
Organization Name
Inventor(s)
APPARATUS AND METHOD WITH NEURAL NETWORK IMPLEMENTATION OF DOMAIN ADAPTATION - A simplified explanation of the abstract
This abstract first appeared for US patent application 18103887 titled 'APPARATUS AND METHOD WITH NEURAL NETWORK IMPLEMENTATION OF DOMAIN ADAPTATION
Simplified Explanation
The abstract describes a method for operating a neural network that involves domain adaptation and style matching. Here is a simplified explanation of the abstract:
- The method involves using a pre-trained neural network and a style feature from a source domain.
- Input data from a target domain is received and a style feature specific to the target domain is extracted.
- The neural network is used to perform domain adaptation on the input data by matching the styles of the source and target domains.
- The style-matched input data is then processed using the neural network.
Potential applications of this technology:
- Image recognition and classification: The method can be used to adapt a pre-trained neural network to different styles of images, allowing for accurate recognition and classification in various domains.
- Natural language processing: The method can be applied to adapt a language model to different writing styles or domains, improving the accuracy of tasks like sentiment analysis or text generation.
Problems solved by this technology:
- Domain adaptation: The method addresses the challenge of adapting a neural network trained on one domain to perform well on a different domain.
- Style matching: By matching the styles of the source and target domains, the method ensures that the neural network can effectively process input data from the target domain.
Benefits of this technology:
- Improved accuracy: The domain adaptation and style matching techniques enhance the performance of the neural network on input data from different domains, leading to more accurate results.
- Flexibility: The method allows for the adaptation of a pre-trained neural network to various domains without the need for extensive retraining, saving time and computational resources.
Original Abstract Submitted
A processor-implemented neural network operating method, the operating method comprising obtaining a neural network pre-trained in a source domain and a first style feature of the source domain, extracting a second style feature of a target domain from received input data of the target domain, using the neural network, performing domain adaptation of the input data, by performing style matching of the input data based on the first style feature of the source domain and the second style feature of the target domain, and processing the style-matched input data, using the neural network.
