17738931. SYSTEM AND METHOD FOR TRAINING A NEURAL NETWORK UNDER PERFORMANCE AND HARDWARE CONSTRAINTS simplified abstract (Samsung Electronics Co., Ltd.)
Contents
SYSTEM AND METHOD FOR TRAINING A NEURAL NETWORK UNDER PERFORMANCE AND HARDWARE CONSTRAINTS
Organization Name
Inventor(s)
Jun Fang of Santa Clara CA (US)
David Philip Lloyd Thorsley of Morgan Hill CA (US)
Joseph H. Hassoun of Los Gatos CA (US)
Hamzah Ahmed Ali Abdelaziz of San Jose CA (US)
SYSTEM AND METHOD FOR TRAINING A NEURAL NETWORK UNDER PERFORMANCE AND HARDWARE CONSTRAINTS - A simplified explanation of the abstract
This abstract first appeared for US patent application 17738931 titled 'SYSTEM AND METHOD FOR TRAINING A NEURAL NETWORK UNDER PERFORMANCE AND HARDWARE CONSTRAINTS
Simplified Explanation
The abstract describes a system and method for training a neural network. It involves training a full-sized network and multiple sub-networks through supervised co-training iterations. Each iteration includes co-training the full-sized network with a subset of the sub-networks.
- The method involves training a full-sized network and multiple sub-networks.
- Supervised co-training is performed through multiple iterations.
- Each iteration includes co-training the full-sized network and a subset of the sub-networks.
Potential Applications
This technology has potential applications in various fields, including:
- Artificial intelligence
- Machine learning
- Pattern recognition
- Data analysis
Problems Solved
The system and method addressed the following problems:
- Training a neural network efficiently
- Enhancing the performance of the neural network
- Improving the accuracy of the network's predictions
Benefits
The benefits of this technology include:
- Improved training efficiency
- Enhanced performance of the neural network
- Increased accuracy in predictions
Original Abstract Submitted
A system and method for training a neural network. In some embodiments the method includes training a full-sized network and a plurality of sub-networks, the training including performing a plurality of iterations of supervised co-training, the performing of each iteration including co-training the full-sized network and a subset of the sub-networks.