Robuta

https://gradientscience.org/robust_apps/ Robustness Beyond Security: Computer Vision Applications – gradient science An off-the-shelf robust classifier can be used to perform a range of computer vision tasks beyond classification. computer vision applications gradient science robustness beyond security https://gradientscience.org/modelcomponents-editing/ Editing Predictions by Modeling Model Computation – gradient science We use our component modeling framework to design targeted model edits. gradient science editing predictions modeling computation https://gradientscience.org/ gradient science Research highlights and perspectives on machine learning and optimization from MadryLab. gradient science https://gradientscience.org/modelcomponents/ Decomposing Predictions by Modeling Model Computation – gradient science We introduce a framework called component modeling for studying how model components collectively shape ML predictions. gradient science predictions modeling computation https://gradientscience.org/adv/ Adversarial Examples Are Not Bugs, They Are Features – gradient science A new perspective on adversarial perturbations adversarial examples are not gradient science bugs features https://gradientscience.org/photoguard/ Raising the Cost of Malicious AI-Powered Image Editing – gradient science Inspired by an episode of the Daily Show, we hacked together a technique for the cost ai powered image editing gradient science raising https://gradientscience.org/batchnorm/ How does Batch Normalization Help Optimization? – gradient science An closer look at one of the most popular techniques in modern deep learning. how does batch normalization gradient science help optimization https://gradientscience.org/unadversarial/ Unadversarial Examples: Designing Objects for Robust Vision – gradient science We show how to design objects to help, rather than hurt, the performance of vision systems; the resulting objects improve performance on natural and... gradient science examples designing objects robust https://gradientscience.org/datamodels-2/ Uncovering Brittleness with Datamodels – gradient science In the second part of our datamodels series, we use datamodels to identify and study a new form of model brittleness. gradient science uncovering brittleness