18467457. METHOD AND DEVICE WITH TRAINING DATABASE CONSTRUCTION simplified abstract (SAMSUNG ELECTRONICS CO., LTD.)
METHOD AND DEVICE WITH TRAINING DATABASE CONSTRUCTION
Organization Name
Inventor(s)
Kyunghyun Cho of New York NY (US)
METHOD AND DEVICE WITH TRAINING DATABASE CONSTRUCTION - A simplified explanation of the abstract
This abstract first appeared for US patent application 18467457 titled 'METHOD AND DEVICE WITH TRAINING DATABASE CONSTRUCTION
Simplified Explanation
The patent application describes an electronic device that utilizes a machine learning-based conditional generative model to reconstruct target data from latent vectors and existing data sets. It also involves extrapolating augmented latent vectors and condition data to generate a new dataset with augmented target data.
- Implement machine learning-based conditional generative model
- Reconstruct target data from latent vectors
- Train model based on existing data set for target task
- Determine extrapolation weight
- Generate augmented latent vector and condition data by extrapolating from existing data
- Create new dataset with augmented target data based on model and augmented data
Potential Applications
- Data augmentation in machine learning tasks
- Image and video generation
- Speech synthesis
Problems Solved
- Generating new data from existing datasets
- Improving model performance through data augmentation
Benefits
- Enhanced model training
- Increased dataset diversity
- Improved accuracy in target data reconstruction
Original Abstract Submitted
An electronic device includes one or more processors and a memory storing instructions configured to, when executed by the one or more processors, cause the one or more processors to: implement a machine learning-based conditional generative model configured to reconstruct target data from latent vectors, the conditional generative model trained based on an existing data set for a target task; determine an extrapolation weight; generate an augmented latent vector and augmented condition data by extrapolating, based on the extrapolation weight, from a latent vector corresponding to the existing dataset and from existing condition data corresponding to the existing dataset; and generate a new dataset comprising augmented target data generated by the conditional generative model based on the augmented condition data and based on the augmented latent vector.
