Code and hyperparameters for the paper "Generative Adversarial Networks"
Role in this project:
ML Engineer Contributions:13 commits, 1 PR, 1 push in 4 years
Contributions summary:Ian's contributions focused on modifying and extending the code for a Generative Adversarial Network (GAN). They adapted code and hyperparameters, fixed issues with data handling, and worked on likelihood evaluation methods. The changes included alterations to core model components, likelihood calculations, and MNIST-specific implementations, revealing a focus on practical application and refinement of the GAN model.
generative-adversarial-networkhyperparameters
An adversarial example library for constructing attacks, building defenses, and benchmarking both
Role in this project:
ML Engineer Contributions:1 release, 918 commits, 503 PRs in 2 years 6 months
Contributions summary:Ian contributed code copied from another repository, adding and modifying TensorFlow-based utilities for constructing and evaluating adversarial examples within the CleverHans framework. The contributions focused on implementing and training models and developing techniques for machine learning security. The changes included adding a code of conduct and implementing a fix for NotImplemtedError to support and maintain CleverHans framework.
benchmarkingmachine-learningsecurity