robocrunch
Ran Zmigrod
@RanZmigrod
Computer Science PhD Candidate at @Cambridge_Uni. Specialising in NLP and algorithms
Tweets by Ran Zmigrod
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
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
Shared by
Ran Zmigrod
at
5/5/2022
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
The K-best algorithm appears to be “feared” in dep parsing (as reviewer 1 called it). Our paper simplifies the base alg (due to Camerini et al. (1980)). In addition to being simpler, it’s 1.4x faster & has a considerably shorter proof of correctness (3 vs 10 pages).
Shared by
Ran Zmigrod
at
6/4/2021
To appear in #ACL2021NLP On Finding the K-best Non-projective Dependency Trees https://t.co/iafgV1V6rA 🌳Simplifies "one of the most feared algorithms” in the dep parsing literature -- Reviewer 1 🌴Produces single-root trees req'd by UD guidelines w/ @xtimv & @ryandcotterell
Shared by
Ran Zmigrod
at
6/4/2021