https://dl.acm.org/doi/10.5555/3305381.3305576. A Survey of Methods for Explaining Black Box Models I recommend you to change the following parameters to your liking. The final report is due April 7. To scale up influence functions to modern [] Imagenet classification with deep convolutional neural networks. Why Use Influence Functions? A sign-up sheet will be distributed via email. We'll cover first-order Taylor approximations (gradients, directional derivatives) and second-order approximations (Hessian) for neural nets. Insights from a noisy quadratic model. On the importance of initialization and momentum in deep learning. Deep inside convolutional networks: Visualising image classification models and saliency maps. we develop a simple, efficient implementation that requires only oracle access to gradients Understanding Black-box Predictions via Influence Functions by Pang Wei Koh and Percy Liang. The dict structure looks similiar to this: Harmful is a list of numbers, which are the IDs of the training data samples The next figure shows the same but for a different model, DenseNet-100/12. The previous lecture treated stochasticity as a curse; this one treats it as a blessing. $-hm`nrurh%\L(0j/hM4/AO*V8z=./hQ-X=g(0
/f83aIF'Mu2?ju]n|# =7$_--($+{=?bvzBU[.Q. Visualised, the output can look like this: The test image on the top left is test image for which the influences were Datta, A., Sen, S., and Zick, Y. Algorithmic transparency via quantitative input influence: Theory and experiments with learning systems. C. Maddison, D. Paulin, Y.-W. Teh, B. O'Donoghue, and A. Doucet. We show that even on non-convex and non-differentiable models where the theory breaks down, approximations to influence functions can still provide valuable information. >> The datasets for the experiments can also be found at the Codalab link. We'll consider two models of stochastic optimization which make vastly different predictions about convergence behavior: the noisy quadratic model, and the interpolation regime. In, Moosavi-Dezfooli, S., Fawzi, A., and Frossard, P. Deep-fool: a simple and accurate method to fool deep neural networks. In order to have any hope of understanding the solutions it comes up with, we need to understand the problems. Therefore, if we bring in an idea from optimization, we need to think not just about whether it will minimize a cost function faster, but also whether it does it in a way that's conducive to generalization. fast SSD, lots of free storage space, and want to calculate the influences on Understanding Black-box Predictions via Influence Functions International Conference on Machine Learning (ICML), 2017. Model selection in kernel based regression using the influence function. Overwhelmed? 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. Natural gradient works efficiently in learning. We look at what additional failures can arise in the multi-agent setting, such as rotation dynamics, and ways to deal with them.