Ran Zmigrod: Tired of approximating the p-value o...| Robocrunch AI

Tired of approximating the p-value of the paired-permutation test? Our new paper gives an 𝑒𝑥𝑎𝑐𝑡 algorithm for a common class of paired-permutation tests that is 𝑓𝑎𝑠𝑡𝑒𝑟 than Monte Carlo approximation!https://t.co/Px7lhUhHmF

with @xtimv @ryandcotterell at @naaclmeeting pic.twitter.com/qM7COxYpqH
— Ran Zmigrod (@RanZmigrod) May 5, 2022

Tired of approximating the p-value of the paired-permutation test? Our new paper gives an 𝑒𝑥𝑎𝑐𝑡 algorithm for a common class of paired-permutation tests that is 𝑓𝑎𝑠𝑡𝑒𝑟 than Monte Carlo approximation! https://t.co/Px7lhUhHmF with @timv @ryandcotterell at @naaclmeeting

Shared by Ran Zmigrod at 5/4/2022

We propose a class of additively structured test statistics (e.g., accuracy) for which the paired-permutation test on a dataset of size N can be reduced to the N-fold convolution of weighted uniform PMFs. Using the FFT, we can do the computation in O(N (log N) (log N)) time.

Shared by Ran Zmigrod at 5/5/2022

@PetarV_93 is right that folks are misinterpreting the math 1. I love the concept permutation-compatibility 🏆 2. But the paper seems to equate the GNN algorithm with equivariance. These are different. See our Equivalence paper to understand it 🧵

Shared by Bruno Ribeiro at 3/10/2022

Repeated experiments guided by sequential model optimization quickly hone in on highly synergistic drug combinations. Mila's Paul Bertin, Yoshua Bengio & @RelationRx Co-founder @JPTK_TechBio shed light on the new RECOVER results in latest blog post. https://t.co/beeq26tARB

Shared by MilaQuebec at 3/23/2022

MMDetection 2.24.0 supports Simple Copy Paste (https://t.co/C1WOUruqJ9) with fully-reproduced performance! 😍Seems many interesting training samples are created. 😆 #AI #Algorithm #PyTorch Welcome to try at: https://t.co/rZPl4H1ykm

Shared by OpenMMLab at 4/26/2022

You also might want to check out our #ICLR2022 paper “Generalization of Neural Combinatorial Solvers Through the Lens of Adversarial Robustness” https://t.co/myIrf4t3Av (joint work with @geisler_si @j_m_sommer @SchuchardtJan @abojchevski)

Shared by Stephan Günnemann at 3/5/2022

768