3 papers
cs.LG2024
Minimizing Chebyshev Prototype Risk Magically Mitigates the Perils of Overfitting
Nathaniel Dean, Dilip Sarkar
Overparameterized deep neural networks (DNNs), if not sufficiently regularized, are susceptible to overfitting their training examples and not generalizing well to test data. To di…
cs.LG2023
Fantastic DNN Classifiers and How to Identify them without Data
Nathaniel Dean, Dilip Sarkar
Current algorithms and architecture can create excellent DNN classifier models from example data. In general, larger training datasets result in better model estimations, which imp…
cs.CV2022
A Perturbation Resistant Transformation and Classification System for Deep Neural Networks
Nathaniel Dean, Dilip Sarkar
Deep convolutional neural networks accurately classify a diverse range of natural images, but may be easily deceived when designed, imperceptible perturbations are embedded in the…