ANanishathalye·Mar 29, 2021
Also of interest might be cleanlab (https://github.com/cgnorthcutt/cleanlab), the open-source software we used for initially identifying potentially mislabeled data.
In Show HN: Label errors in benchmark ML test setscgnorthcutt
cleanlab
OPEN SOURCE2019–2021
cgnorthcutt / cleanlab
cleanlab
Official cleanlab repo is at https://github.com/cleanlab/cleanlab
2Distinct people
2Mentions
0Explicit recommenders
57GitHub stars
What to know before exploring
What it is
Official cleanlab repo is at https://github.com/cleanlab/cleanlab
GitHub
Where it is used
No documented product use has been verified yet.
What people are saying
No recommendation phrase was found in the indexed comments.
No classified warning is attached to this page. Read the discussions before treating silence as approval.
Also in these discussions
No other repository is repeatedly linked in the same discussions.
This week on Hacker News
Nov 7–Nov 14 UTC · Direct repository links and separately labeled related project stories. Indexed through 11/14/2026, 3:56:07 PM UTC.
No direct repository links were found in the indexed HN discussions over the past seven days. Older discussions are available below.
Mentions over time
HN mentions by month2 dated mentions
Observed mentions: 2019-11-04 14:29:39 UTC to 2021-03-29 16:32:39 UTC.
Counts come from the repository’s dated source evidence. Index cutoff: 2026-11-02 08:16:59 UTC.
View mention counts and date ranges
| Period | Start (UTC) | End (exclusive, UTC) | Mentions | Submissions | Comments |
|---|---|---|---|---|---|
| Nov 2019 | 2019-11-01 00:00:00 UTC | 2019-12-01 00:00:00 UTC | 1 | 0 | 1 |
| Dec 2019 | 2019-12-01 00:00:00 UTC | 2020-01-01 00:00:00 UTC | 0 | 0 | 0 |
| Jan 2020 | 2020-01-01 00:00:00 UTC | 2020-02-01 00:00:00 UTC | 0 | 0 | 0 |
| Feb 2020 | 2020-02-01 00:00:00 UTC | 2020-03-01 00:00:00 UTC | 0 | 0 | 0 |
| Mar 2020 | 2020-03-01 00:00:00 UTC | 2020-04-01 00:00:00 UTC | 0 | 0 | 0 |
| Apr 2020 | 2020-04-01 00:00:00 UTC | 2020-05-01 00:00:00 UTC | 0 | 0 | 0 |
| May 2020 | 2020-05-01 00:00:00 UTC | 2020-06-01 00:00:00 UTC | 0 | 0 | 0 |
| Jun 2020 | 2020-06-01 00:00:00 UTC | 2020-07-01 00:00:00 UTC | 0 | 0 | 0 |
| Jul 2020 | 2020-07-01 00:00:00 UTC | 2020-08-01 00:00:00 UTC | 0 | 0 | 0 |
| Aug 2020 | 2020-08-01 00:00:00 UTC | 2020-09-01 00:00:00 UTC | 0 | 0 | 0 |
| Sep 2020 | 2020-09-01 00:00:00 UTC | 2020-10-01 00:00:00 UTC | 0 | 0 | 0 |
| Oct 2020 | 2020-10-01 00:00:00 UTC | 2020-11-01 00:00:00 UTC | 0 | 0 | 0 |
| Nov 2020 | 2020-11-01 00:00:00 UTC | 2020-12-01 00:00:00 UTC | 0 | 0 | 0 |
| Dec 2020 | 2020-12-01 00:00:00 UTC | 2021-01-01 00:00:00 UTC | 0 | 0 | 0 |
| Jan 2021 | 2021-01-01 00:00:00 UTC | 2021-02-01 00:00:00 UTC | 0 | 0 | 0 |
| Feb 2021 | 2021-02-01 00:00:00 UTC | 2021-03-01 00:00:00 UTC | 0 | 0 | 0 |
| Mar 2021 | 2021-03-01 00:00:00 UTC | 2021-04-01 00:00:00 UTC | 1 | 0 | 1 |
What Hacker News says
2 sourcesHi, Hackers. I'm excited to share confident learning for characterizing, finding, and learning with label errors in datasets. To promote and standardize future research in learning with noisy labels and weak supervision, I've also open-sourced the cleanlab Python package: https://pypi.org/project/cleanlab/
Post: https://l7.curtisnorthcutt.com/confident-learning
Paper: https://arxiv.org/abs/1911.00068
Code: https://github.com/cgnorthcutt/cleanlab/
Abstract: Learning exists in the context of data, yet notions of confidence typically focus on model predictions, not label quality. Confident learning (CL) has emerged as an approach for characterizing, identifying, and learning with noisy labels in datasets, based on the principles of pruning noisy data, counting to estimate noise, and ranking examples to train with confidence. Here, we general
Read the rest
ize CL, building on the assumption of a classification noise process, to directly estimate the joint distribution between noisy (given) labels and uncorrupted (unknown) labels. This generalized CL, open-sourced as cleanlab, is provably consistent under reasonable conditions, and experimentally performant on ImageNet and CIFAR, outperforming recent approaches, e.g. MentorNet, by 30% or more, when label noise is non-uniform. cleanlab also quantifies ontological class overlap, and can increase model accuracy (e.g. ResNet) by providing clean data for training.