International business machines corporation (20240289623). EDITING A TARGET MODEL TO FORGET DATA SAMPLES USING A REFERENCE MODEL TO ADJUST WEIGHTS OF THE TARGET MODEL simplified abstract
Contents
- 1 EDITING A TARGET MODEL TO FORGET DATA SAMPLES USING A REFERENCE MODEL TO ADJUST WEIGHTS OF THE TARGET MODEL
- 1.1 Organization Name
- 1.2 Inventor(s)
- 1.3 EDITING A TARGET MODEL TO FORGET DATA SAMPLES USING A REFERENCE MODEL TO ADJUST WEIGHTS OF THE TARGET MODEL - A simplified explanation of the abstract
- 1.4 Simplified Explanation
- 1.5 Potential Applications
- 1.6 Problems Solved
- 1.7 Benefits
- 1.8 Commercial Applications
- 1.9 Prior Art
- 1.10 Frequently Updated Research
- 1.11 Questions about Data Privacy Enhancement Tool
- 1.12 Original Abstract Submitted
EDITING A TARGET MODEL TO FORGET DATA SAMPLES USING A REFERENCE MODEL TO ADJUST WEIGHTS OF THE TARGET MODEL
Organization Name
international business machines corporation
Inventor(s)
Ron Shmelkin of Givatayim (IL)
Abigail Goldsteen of Haifa (IL)
Ariel Farkash of Shimshit (IL)
EDITING A TARGET MODEL TO FORGET DATA SAMPLES USING A REFERENCE MODEL TO ADJUST WEIGHTS OF THE TARGET MODEL - A simplified explanation of the abstract
This abstract first appeared for US patent application 20240289623 titled 'EDITING A TARGET MODEL TO FORGET DATA SAMPLES USING A REFERENCE MODEL TO ADJUST WEIGHTS OF THE TARGET MODEL
Simplified Explanation
The patent application describes a computer program product, system, and method for editing a target model to forget data samples. This involves inputting forget data samples into a reference model trained on a non-private data set to produce reference output. The forget data samples are then inputted into a target model trained on a total data set comprising the non-private data set and a private data set to produce target output. Gradients are calculated to minimize the error of a loss function measuring the difference between reference and target outputs, and these gradients are used to update weights in the target model.
- Forget data samples are inputted into a reference model trained on a non-private data set.
- The forget data samples are then inputted into a target model trained on a total data set including a private data set.
- Gradients are calculated to minimize the error of a loss function measuring the difference between reference and target outputs.
- The calculated gradients are used to update weights in the target model.
- This process results in an edited target model.
Potential Applications
This technology can be applied in various fields such as machine learning, data privacy, and artificial intelligence.
Problems Solved
This technology addresses the need to selectively forget specific data samples in a target model while maintaining overall model performance.
Benefits
The benefits of this technology include improved data privacy, enhanced model flexibility, and the ability to adapt to changing data requirements.
Commercial Applications
Title: Data Privacy Enhancement Tool This technology can be used in industries where data privacy is crucial, such as healthcare, finance, and cybersecurity. It can help organizations comply with data protection regulations and enhance the security of sensitive information.
Prior Art
Prior research in machine learning and data privacy may provide insights into similar techniques for forgetting data samples in models.
Frequently Updated Research
Researchers are continually exploring new methods for enhancing data privacy and model performance in machine learning systems. Stay updated on recent advancements in this field for the latest innovations.
Questions about Data Privacy Enhancement Tool
1. How does this technology ensure the privacy of sensitive data during the model editing process? 2. What are the potential implications of using this technology in industries with strict data privacy regulations?
Original Abstract Submitted
provided are a computer program product, system, and method for editing a target model to forget data samples. forget data samples of data samples to forget are inputted into a reference model, trained on a non-private data set, to produce reference output. the forget data samples to forget are inputted to a target model, trained on a total data set comprising the non-private data set and a private data set, to produce target output. the private data set includes the forget data samples a loss function is calculated to measure a divergence of the reference output and the target output. a determination is made of gradients that minimize an error of the loss function. optimized gradients are calculated from the determined gradients. the optimized gradients are applied to update weights in the target model to produce an edited target model.
