17849391. TOKEN SYNTHESIS FOR MACHINE LEARNING MODELS simplified abstract (Capital One Services, LLC)
Contents
TOKEN SYNTHESIS FOR MACHINE LEARNING MODELS
Organization Name
Inventor(s)
Anh Truong of Champaign IL (US)
Jeremy Goodsitt of Champaign IL (US)
Austin Walters of Savoy IL (US)
TOKEN SYNTHESIS FOR MACHINE LEARNING MODELS - A simplified explanation of the abstract
This abstract first appeared for US patent application 17849391 titled 'TOKEN SYNTHESIS FOR MACHINE LEARNING MODELS
Simplified Explanation
The abstract describes a method for generating vectors associated with tokens using learning models and clustering criteria. It also involves modifying the vectors to generate perturbed vectors and synthesizing tokens based on these perturbed vectors. A second learning model is then trained based on the synthesized tokens.
- Method for generating vectors associated with tokens using learning models and clustering criteria
- Modifying vectors to generate perturbed vectors associated with a different label
- Updating vector sets with the perturbed vectors
- Synthesizing tokens based on the perturbed vectors using a different set of parameters
- Training a second learning model based on the synthesized tokens
Potential Applications
- Natural language processing
- Sentiment analysis
- Text classification
- Information retrieval
Problems Solved
- Improving the accuracy of learning models by generating perturbed vectors
- Enhancing the performance of clustering algorithms
- Increasing the efficiency of training learning models
Benefits
- Improved accuracy and performance of learning models
- Enhanced clustering algorithms
- More efficient training of learning models
Original Abstract Submitted
A method includes generating a first vector associated with a token using a first set of parameters of a first learning model based on the token, determining a prediction indicating that the token is associated with a first label based on a set of clustering criteria, the first vector, vectors of a first vector set, and vectors of a second vector. The method includes generating a perturbed vector associated with a second label by modifying a value of the first vector and updating the second vector set to comprise the perturbed vector. The method also includes generating a synthesized token associated with the second vector set based on the perturbed vector using a second set of parameters of the first learning model and training a second learning model based on the synthesized token.
