Colleen Chen

@HowitzerJ @CELRL_Melbourne @Government_UoM Thanks for your guidance on this @HowitzerJ

Very pleased to have collaborated with @colleen_chen_ on this new @CELRL_Melbourne Policy Brief calling for a Worker Data Right to give workers more control of their data @Government_UoM #RegulationandDesign

(1/12) I am excited to share the #NLProc project we've been working on for the past year: #MultiLexSum A multi-doc #summarization dataset for civil rights lawsuits.🌟It has summaries of 3 different granularities! Website: multilexsum.github.io Paper: arxiv.org/abs/2206.10883 🧵

@BrentHodgson @TimothyKariotis

VICTORIANS: What do you think our state politicians can do better? And what are the issues you’d like to see addressed in the next parliament?

This website creates satellite images of cities that don't exist! AI generated satellite imagery Every time you refresh the page a new image of city that does not exist is created. What could possibly go wrong 🤔 thiscitydoesnotexist.com

New (to me) dimension of crowdwork platforms: Russian military uses Premise microtasking platform to aim and calibrate fire during their invasion of Ukraine. Example tasks are to locate ports, medical facilities, bridges, explosion craters. Paying ¢0.25 to $3.25 a task.

Any seismologists out there that might be interested in helping create a podcast episode around mapping earthquakes and using earthquakes to map the earth? I would love to hear from you!

From mapping poverty hotspots to digitising languages, a growing number of African developers are using AI to tackle their continent's particular challenges 🌍 @bonjora @DeepIndaba @ulrichpaquet @BenjaminRosman @cinalawson @NovissiTG @MasakhaneNLP news.trust.org/item/202202161…

First map ever of "dark fishing" (vessels without AIS) in the Mediterranean. One year's of vessel detections from Sentinel-1, matched to AIS, @GlobalFishWatch: globalfishingwatch.org/data/radar-ill… (orange is without AIS, green is with)

How do models like GPT-2 and BERT represent position of tokens? When visualizing their positional encodings I found an interesting pattern. A short thread:

When a 'machine' mentioned in a 19th century newspaper article could be a pram, Parliament, a lever, a sewing machine or a new type of loom, understanding the impact of machines on ordinary lives is... tricky. Help @LivingWMachines researchers today! zooniverse.org/projects/bldig…

RIP Leland Wilkinson. He changed the #dataviz game forever with 'Grammar of Graphics' (the "gg" in ggplot), he said things like this that made me laugh and that made a lot of sense to me. While I didn't know him well, he showed me kindness once that had a big impact on me.

✨✨New Package✨✨ Happy to announce that I am soft launching my R color palette package {MetBrewer} today! Currently has over 30 palettes inspired by works at the @metmuseum and expanding! Download Instruction and Palettes here: bit.ly/3GDIj1N #r4ds #dataviz #rstats

Descriptions, captions, and alt text can help make visualizations more accessible. But what should they say, or not say? My new paper with @arvindsatya1 contributes a model for evaluating textual descriptions of charts. Accessible HTML Paper: vis.csail.mit.edu/pubs/vis-text-… #a11y /Thread

Existing stock photos for Artificial Intelligence display shiny humanoid robots, glowing blue brains or the Terminator. These are unhelpful stereotypes; they set unrealistic expectations and hinder a wider understanding of the technology and its societal and environmental impact

Important work is happening in *Participatory ML* and in recognizing that AI ethics is about the *distribution of power*. I want to create a thread linking to some of this work 1/ twitter.com/math_rachel/st…

New preprint: Policymakers worldwide call for "human oversight" to mitigate the risks of government algorithms. Yet rather than protect against algorithmic harms, human oversight legitimizes flawed and unaccountable government algorithms. 📑: papers.ssrn.com/sol3/papers.cf…

The use of algorithms has been decried for entrenching discrimination and injustices against people of color in the criminal legal system. But @ColinDoyleLaw writes that algorithms, instead, can be unlikely allies in the fight against mass incarceration. inquest.org/the-feature-is…

To better understand the ethics of machine learning datasets, we picked three controversial face recognition / person recognition datasets—DukeMTMC, MS-Celeb-1M, and Labeled Faces in the Wild—and analyzed ~1,000 papers that cite them. Paper: arxiv.org/pdf/2108.02922… Thread ⬇️

Can machine learning outperform baseline logistic regression for predicting complex social phenomena? Many prominent papers have claimed highly accurate civil war prediction. In a systematic review, @sayashk and I find these claims invalid due to errors. reproducible.cs.princeton.edu