Systems and methods for generating datasets for model retraining

A method implemented on a computer system, the method comprising:

• storing, to a non-transitory storage medium, a reference model and a reference dataset that is associated with the reference model;
• generating a plurality of synthetic datasets that are derived from detection events that are labeled;
• generating, for each respective synthetic dataset of the plurality of synthetic datasets, a plurality of feature metrics for a plurality of features of each respective synthetic dataset;
• for each respective one of the plurality of synthetic datasets, using a first neural network to generate, based on the plurality of feature metrics for the respective one of the plurality of synthetic datasets, a dataset similarity score with respect to the reference dataset, wherein each dataset similarity score for each respective one of the plurality of synthetic datasets indicates how similar the respective one is to the reference dataset;
• generating, for each of the plurality of synthetic datasets, a training similarity score by training a neural network architecture by using a corresponding synthetic dataset;
• and
• constructing a synthesized dataset by blending the plurality of synthetic datasets by using generated training similarity scores and the dataset similarity scores.

A non-transitory computer readable storage medium configured to store computer-executable instructions for use with a computer system, the stored computer-executable instructions comprising instructions that cause the computer system to perform operations comprising:

• storing, to a non-transitory storage medium, a reference model and a reference dataset that is associated with the reference model;
• generating a plurality of synthetic datasets that are derived from labeled detection events;
• generating, for each respective synthetic dataset of the plurality of synthetic datasets, a plurality of feature metrics for a plurality of features of each respective synthetic dataset;
• for each respective one of the plurality of synthetic datasets, using a first neural network to generate, based on the plurality of feature metrics for the respective one of the plurality of synthetic datasets, a dataset similarity score with respect to the reference dataset, wherein each dataset similarity score for each respective one of the plurality of synthetic datasets indicates how similar the respective one is to the reference dataset;
• generating, for each of the plurality of synthetic datasets, a training similarity score by training a second neural network by using a corresponding synthetic dataset;
• and
• constructing a synthesized dataset by blending the plurality of synthetic datasets by using generated training similarity scores and the dataset similarity scores.