18542133. METHODS, SYSTEMS, APPARATUS, AND ARTICLES OF MANUFACTURE TO AUGMENT TRAINING DATA BASED ON SYNTHETIC IMAGES simplified abstract (Intel Corporation)
METHODS, SYSTEMS, APPARATUS, AND ARTICLES OF MANUFACTURE TO AUGMENT TRAINING DATA BASED ON SYNTHETIC IMAGES
Organization Name
Inventor(s)
Anmol Bhasin of SAS Nagar (IN)
Shekar Ramachandran of Bengaluru (IN)
Rudra Nath Palit of Kolkata (IN)
Rupali Agrahari of Sultanpur (IN)
Sai Pramod Gadam of Bengaluru (IN)
METHODS, SYSTEMS, APPARATUS, AND ARTICLES OF MANUFACTURE TO AUGMENT TRAINING DATA BASED ON SYNTHETIC IMAGES - A simplified explanation of the abstract
This abstract first appeared for US patent application 18542133 titled 'METHODS, SYSTEMS, APPARATUS, AND ARTICLES OF MANUFACTURE TO AUGMENT TRAINING DATA BASED ON SYNTHETIC IMAGES
Simplified Explanation
The patent application describes methods, systems, and apparatus for augmenting training data based on synthetic images using a generative adversarial network (GAN). Here is a simplified explanation of the patent application:
- Programmable circuitry generates a latent representation of an image from one racial domain and then generates a new image from a different racial domain to augment a training dataset.
Potential Applications:
- Image generation for training datasets in various industries such as computer vision, artificial intelligence, and machine learning.
Problems Solved:
- Lack of diverse training data for machine learning models.
- Difficulty in generating synthetic images that accurately represent different racial domains.
Benefits:
- Improved accuracy and performance of machine learning models.
- Increased diversity and representation in training datasets.
Potential Commercial Applications:
- AI and machine learning companies can use this technology to enhance their training datasets.
- Companies in industries like healthcare, finance, and retail can benefit from more accurate and diverse data for their AI applications.
Possible Prior Art:
- Prior art may include similar methods for generating synthetic images for training datasets using GANs.
Unanswered Questions: 1. How does the patent application address potential biases in the generated synthetic images? 2. What are the computational requirements for implementing this technology at scale in real-world applications?
Original Abstract Submitted
Methods, systems, apparatus, and articles of manufacture to augment training data based on synthetic images are disclosed. An example apparatus disclosed herein includes programmable circuitry to generate, with one or more first layers of a generative adversarial network (GAN), a latent representation corresponding to a first image representative of a first racial domain, generate, with one or more second layers of the GAN, a second image based on the latent representation, the second image corresponding to a second racial domain different from the first racial domain, and augment a training dataset based on the second image.
