Difference between revisions of "18045746. MONITORING MACHINE LEARNING MODELS USING SURROGATE MODEL OUTPUT simplified abstract (Capital One Services, LLC)"
Wikipatents (talk | contribs) (Creating a new page) |
(No difference)
|
Latest revision as of 03:58, 16 April 2024
Contents
- 1 MONITORING MACHINE LEARNING MODELS USING SURROGATE MODEL OUTPUT
- 1.1 Organization Name
- 1.2 Inventor(s)
- 1.3 MONITORING MACHINE LEARNING MODELS USING SURROGATE MODEL OUTPUT - A simplified explanation of the abstract
- 1.4 Simplified Explanation
- 1.5 Potential Applications
- 1.6 Problems Solved
- 1.7 Benefits
- 1.8 Potential Commercial Applications
- 1.9 Possible Prior Art
- 1.10 Original Abstract Submitted
MONITORING MACHINE LEARNING MODELS USING SURROGATE MODEL OUTPUT
Organization Name
Inventor(s)
Samuel Sharpe of Cambridge MA (US)
Christopher Bayan Bruss of Washington DC (US)
Brian Barr of Schenectady NY (US)
Justin Au-yeung of Somerville MA (US)
MONITORING MACHINE LEARNING MODELS USING SURROGATE MODEL OUTPUT - A simplified explanation of the abstract
This abstract first appeared for US patent application 18045746 titled 'MONITORING MACHINE LEARNING MODELS USING SURROGATE MODEL OUTPUT
Simplified Explanation
The abstract describes a patent application for a computing system that uses a surrogate machine learning model to detect biases in a production machine learning model based on subpopulations of data. The surrogate model is trained using features not included in the production model's data, such as demographic information, to identify if the production model treats certain subpopulations differently.
- The patent application involves using a surrogate machine learning model to detect biases in a production machine learning model.
- The surrogate model is trained on features not included in the production model's data, such as demographic information.
- The surrogate model helps identify if the production model treats different subpopulations of data differently.
Potential Applications
This technology could be applied in various fields where machine learning models are used to make decisions, such as finance, healthcare, and criminal justice.
Problems Solved
This technology helps identify and mitigate biases in machine learning models, ensuring fair and accurate decision-making processes.
Benefits
The use of a surrogate model can improve the transparency and accountability of machine learning systems, leading to more equitable outcomes for different subpopulations.
Potential Commercial Applications
One potential commercial application of this technology could be in the development of tools for auditing and monitoring machine learning models for bias in industries like banking and insurance.
Possible Prior Art
Prior art in this field may include research on bias detection and mitigation in machine learning models, as well as studies on the impact of demographic information on model performance.
Unanswered Questions
How does the surrogate model compare to other methods for detecting biases in machine learning models?
The article does not provide a comparison with other methods for bias detection, such as fairness metrics or adversarial attacks.
What are the potential limitations or challenges of implementing this technology in real-world applications?
The article does not address the potential challenges of integrating the surrogate model into existing machine learning pipelines or the computational resources required for training and deploying the model.
Original Abstract Submitted
In some aspects, a computing system may use a surrogate machine learning model to detect whether a production or other machine learning model has a tendency to generate different output depending on which subpopulation a particular sample belongs to. The surrogate machine learning model may be trained using features/outputs that are not included in the data used by the production model. For example, by using demographic information in lieu of the original labels of a dataset that was used to train a production model, a surrogate model may be used to detect whether the production model is able to discern one or more characteristics associated with but not present in a sample using other features of the dataset. Output of the surrogate machine learning model may be clustered to detect whether certain subpopulations are treated differently by the production model.