Why Use Influence Functions? as long as you have a supervised learning problem. More details can be found in the project handout. Are you sure you want to create this branch? The project proposal is due on Feb 17, and is primarily a way for us to give you feedback on your project idea. Deep learning via Hessian-free optimization. Understanding Black-box Predictions via Influence Functions by Pang Wei Koh and Percy Liang. In, Metsis, V., Androutsopoulos, I., and Paliouras, G. Spam filtering with naive Bayes - which naive Bayes? Google Scholar Digital Library; Josua Krause, Adam Perer, and Kenney Ng. Borys Bryndak, Sergio Casas, and Sean Segal. Liu, Y., Jiang, S., and Liao, S. Efficient approximation of cross-validation for kernel methods using Bouligand influence function. 2172: 2017: . Optimizing neural networks with Kronecker-factored approximate curvature. Lectures will be delivered synchronously via Zoom, and recorded for asynchronous viewing by enrolled students. Understanding black-box predictions via influence functions. For a point z and parameters 2 , let L(z; ) be the loss, and let1 n P n i=1L(z above, keeping the grad_zs only makes sense if they can be loaded faster/ The datasets for the experiments can also be found at the Codalab link. place. % Noisy natural gradient as variational inference. Apparently this worked. 2019. Understanding short-horizon bias in stochastic meta-optimization. When can we take advantage of parallelism to train neural nets? If you have questions, please contact Pang Wei Koh (pangwei@cs.stanford.edu). test images, the harmfulness is ordered by average harmfullness to the Kelvin Wong, Siva Manivasagam, and Amanjit Singh Kainth. J. Lucas, S. Sun, R. Zemel, and R. Grosse. This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository. How can we explain the predictions of a black-box model? Yuwen Xiong, Andrew Liao, and Jingkang Wang. For modern neural nets, the analysis is more often descriptive: taking the procedures practitioners are already using, and figuring out why they (seem to) work. In this paper, we use influence functions -- a classic technique from robust statistics -- to trace a model's prediction through the learning algorithm and back to its training data, thereby identifying training points most responsible for a given prediction. Besides just getting your networks to train better, another important reason to study neural net training dynamics is that many of our modern architectures are themselves powerful enough to do optimization. The dict structure looks similiar to this: Harmful is a list of numbers, which are the IDs of the training data samples There are various full-featured deep learning frameworks built on top of JAX and designed to resemble other frameworks you might be familiar with, such as PyTorch or Keras. Russakovsky, O., Deng, J., Su, H., Krause, J., Satheesh, S., Ma, S., Huang, Z., Karpathy, A., Khosla, A., Bernstein, M., et al.
I Second The Motion'' Or I Second Demotion, Articles U