Robust and fast model fitting by adaptive sampling

摘要

Aspects of the present disclosure relate generally to model fitting. A target model having a large number of inputs is fit using a performance model having relatively few inputs. The performance model is learned during the fitting process. Optimal optimization parameters including a sample size, a damping factor, and an iteration count are selected for an optimization round. A random subset of data is sampled based on the selected sample size. The optimization round is conducted using the iteration count and the sampled data to produce optimized parameters. The performance model is updated based on the performance of the optimization round. The parameters of the target model are then updated based on the damping factor and the parameters computed by the optimization round. The aforementioned steps are performed in a loop in order to obtain optimized parameters and fit of the data to the target model.

主权项

1. A method for determining a set of optimized target parameters for a target model and training data, the method comprising:
selecting a set of optimization parameters including a sample size, a damping factor, and an iteration value; sampling the training data based on the selected sample size; performing, by a processor, an optimization round including optimizing an objective function for the sampled training data and the selected iteration value; generating a performance model associated with a set of performance parameters based on the optimization round, the selected iteration value, the selected sample size and the damping factor; determining, by the processor, a generalization error for the optimization round; analyzing the generalization error by evaluating the target model on a second sample of the training data; updating the set of performance parameters based on the analysis of the generalization error to obtain a set of updated performance parameters; and updating a set of target parameters based on the performance model, the set of updated performance parameters, and the generalization error to obtain a set of optimized target parameters for the sampled training data.