Leveraging Simple Model Predictions for Enhancing Computational Performance

A system for enhancing predictive performance of a simple model, comprising:

• a processor configured to operate a simple model;
• and
• a memory coupled to the processor, the memory storing instructions to cause the processor to perform acts comprising:
• providing to a complex model a training data set associated with the simple model;
• receiving the training data set and a generated hardness value associated with a prediction of each data point;
• and
• retraining the simple model utilizing a re-weighted training data set.

A non-transitory computer-readable storage medium tangibly embodying a computer-readable program code having computer-readable instructions that, when executed, causes a computer device to perform a method of enhancing performance of a simple model, the method comprising:

• training the simple model with a training data set associated with the simple model;
• receiving, by a complex model, the training data set associated with the simple model;
• generating by the complex model a hardness value associated with a prediction of each data point,
• outputting the training data set and the hardness value to the simple model;
• determining, by the simple model, a confidence value regarding predicting each data point based on the training data set;
• determining a ratio of the hardness value of predicting each data point by the complex model with respect to the confidence value of the simple model;
• and
• retraining the simple model utilizing a re-weighted training data set when the determined ratio is lower than a value of β.