18659457. ELECTRONIC APPARATUS AND METHOD FOR RE-LEARNING TRAINED MODEL simplified abstract (SAMSUNG ELECTRONICS CO., LTD.)
Contents
ELECTRONIC APPARATUS AND METHOD FOR RE-LEARNING TRAINED MODEL
Organization Name
Inventor(s)
ELECTRONIC APPARATUS AND METHOD FOR RE-LEARNING TRAINED MODEL - A simplified explanation of the abstract
This abstract first appeared for US patent application 18659457 titled 'ELECTRONIC APPARATUS AND METHOD FOR RE-LEARNING TRAINED MODEL
- Simplified Explanation:**
The patent application describes a method for re-learning a trained model by selectively re-learning parameters associated with a new task while dynamically expanding the size of the trained model if needed.
- Key Features and Innovation:**
- Re-learning a trained model by identifying neurons associated with a new task and selectively re-learning parameters for those neurons.
- Dynamically expanding the size of the trained model to reconstruct the input model if a preset loss value is reached.
- Enhancing the efficiency of re-learning tasks in neural networks by focusing on specific neurons and parameters.
- Potential Applications:**
This technology could be applied in various fields such as machine learning, artificial intelligence, data analysis, and pattern recognition.
- Problems Solved:**
- Improves the adaptability of trained models to new tasks.
- Enhances the efficiency of re-learning processes in neural networks.
- Allows for targeted parameter adjustments in complex models.
- Benefits:**
- Faster adaptation to new tasks.
- Improved accuracy in re-learning processes.
- Enhanced performance of neural networks in dynamic environments.
- Commercial Applications:**
Potential commercial applications include automated data analysis systems, intelligent decision-making algorithms, and advanced pattern recognition software for various industries.
- Prior Art:**
Prior research in the field of neural network re-training and dynamic model expansion could provide valuable insights into similar approaches.
- Frequently Updated Research:**
Stay updated on the latest advancements in neural network re-training techniques, dynamic model expansion methods, and optimization strategies for complex models.
- Questions about the Technology:**
1. How does the method for selectively re-learning parameters in neural networks improve efficiency? 2. What are the potential implications of dynamically expanding the size of trained models in machine learning applications?
Original Abstract Submitted
A method for re-learning a trained model is provided. The method for re-learning a trained model includes: receiving a data set including the trained model consisting of a plurality of neurons and a new task; identifying a neuron associated with the new task among the plurality of neurons to selectively re-learn a parameter associated with the new task for the identified neuron; and dynamically expanding a size of the trained model on which the selective re-learning is performed if the trained model on which the selective re-learning has a preset loss value to reconstruct the input trained model.