Meta tags:
description=Notes accompanying our ICML tutorial;

Headings (most frequently used words):

data, in, abstract, chapters, problems, and, solution, concepts, ml, ii, theoretical, foundations, iii, scaling, to, deep, learning, iv, attribution, the, wild,

Text of the page (most frequently used words):
data (16), #attribution (13), 2024 (9), the (8), and (7), icml (5), scale (5), this (4), tutorial (4), july (4), chapter (4), with (3), for (3), mit (3), learning (3), predictive (3), problems (2), call (2), one (2), can (2), problem (2), weighted (2), refitting (2), methods (2), directly (2), modern (2), these (2), progress (2), models (2), applications (2), notes (2), site, created, jekyll, tufte, theme, wild, iii, scaling, deep, theoretical, foundations, solution, concepts, chapters, requires, account, coming, soon, youtube, slideslive, video, pdf, slides, surveys, field, focus, what, first, motivate, notion, within, broad, purpose, based, taxonomy, next, highlight, how, view, through, lens, classic, statistical, that, discuss, why, classical, solving, struggle, when, applied, large, machine, settings, thus, cannot, solve, contexts, shortcomings, mind, overview, recent, performing, finally, discussing, current, future, study, relation, between, predictions, downstream, help, interpret, compare, curate, datasets, assess, algorithm, stability, abstract, website, you, will, find, from, were, put, together, they, are, very, much, work, please, feel, free, suggest, edits, improvements, reaching, out, any, emailing, edu, sung, min, sam, park, logan, engstrom, kristian, georgiev, andrew, ilyas, welcome,

Text of the page (random words):
data attribution at scale icml 2024 data attribution at scale chapter 1 chapter 2 chapter 3 chapter 4 welcome on this website you will find notes from the icml 2024 tutorial data attribution at scale these notes were put together by andrew ilyas kristian georgiev logan engstrom and sung min sam park they are very much a work in progress please feel free to suggest edits improvements by reaching out to any one of us or by emailing us at ml data tutorial mit edu abstract data attribution is the study of the relation between data and ml predictions in downstream applications data attribution methods can help interpret and compare models curate datasets and assess learning algorithm stability this tutorial surveys the field of data attribution with a focus on what we call predictive data attribution we first motivate this notion within a broad purpose based taxonomy of data attribution next we highlight how one can view predictive data attribution through the lens of a classic statistical problem that we call weighted refitting we discuss why classical methods for solving the weighted refitting problem struggle when directly applied to large scale machine learning settings and thus cannot directly solve problems in modern contexts with these shortcomings in mind we overview recent progress on performing predictive data attribution for modern ml models finally we discussing applications current and future of data attribution slides pdf video slideslive requires icml account youtube coming soon chapters i data problems and solution concepts in ml july 18 2024 ii theoretical foundations july 18 2024 iii scaling to deep learning july 18 2024 iv data attribution in the wild july 18 2024 data attribution at scale icml 2024 mit 2024 mit this site created with the tufte theme for a tutorial at icml 2024 in jekyll