17808314. BLACK-BOX EXPLAINER FOR TIME SERIES FORECASTING simplified abstract (INTERNATIONAL BUSINESS MACHINES CORPORATION)
Contents
BLACK-BOX EXPLAINER FOR TIME SERIES FORECASTING
Organization Name
INTERNATIONAL BUSINESS MACHINES CORPORATION
Inventor(s)
Vikas C. Raykar of Bangalore (IN)
Sumanta Mukherjee of Bangalore (IN)
Nupur Aggarwal of Bangalore (IN)
Bhanukiran Vinzamuri of Long Island City NY (US)
Arindam Jati of Bengaluru (IN)
BLACK-BOX EXPLAINER FOR TIME SERIES FORECASTING - A simplified explanation of the abstract
This abstract first appeared for US patent application 17808314 titled 'BLACK-BOX EXPLAINER FOR TIME SERIES FORECASTING
Simplified Explanation
The patent application describes a method, system, and computer program for explaining the predictions of univariate time series forecasters that use black-box models. Here are the key points:
- The method receives a set of time series forecasting predictions generated by black-box models trained with an initial data set.
- A set of features is generated based on at least a portion of the initial data set.
- Surrogate models are trained using the set of time series forecasting predictions and the set of features.
- A subset of surrogate models is selected.
- Based on the subset of surrogate models, one or more explanation outputs are generated for the time series forecasting predictions of the black-box models.
Potential applications of this technology:
- Interpreting the predictions of univariate time series forecasters.
- Providing explanations for the predictions made by black-box models.
- Enhancing transparency and trust in the decision-making process of time series forecasting models.
Problems solved by this technology:
- Black-box models often lack interpretability, making it difficult to understand the reasoning behind their predictions.
- Explaining the predictions of time series forecasters can help identify potential biases, errors, or anomalies in the models.
- Providing explanations can improve the adoption and acceptance of time series forecasting models in various domains.
Benefits of this technology:
- Enables users to understand and trust the predictions made by black-box time series forecasters.
- Facilitates the identification of potential issues or limitations in the forecasting models.
- Enhances transparency and accountability in decision-making processes that rely on time series forecasting.
Original Abstract Submitted
A method, system, and computer program product for an interpretable, feature-based post-hoc black box explainer for univariate time series forecasters are provided. The method receives a set of time series forecasting predictions. The set of time series forecasting predictions are generated from a set of black-box models trained with an initial data set. The method generates a set of features based on at least a portion of the initial data set. A set of surrogate models are trained based on the set of time series forecasting predictions and at least a portion of the set of features. A subset of surrogate models is selected. Based on the subset of surrogate models, the method generates one or more explanation outputs for time series forecasting predictions of the set of black-box models.
