Show newer

The #Clarivate list of Highly Cited #Researchers turns out to be a happy hunting ground for identifying dodgy individuals involved in #Papermills and other dubious activities. forbetterscience.com/2022/11/2

As a bonus, this is an European meeting, which makes it possible to travel by train instead of plane (8/7)

Show thread

So we may add a 4th rule for next year's meeting: a short crash course in audiology the week before so that everybody can follow the talks. (7/7)

Show thread

Yet, I realized that most of the questions were asked by senior researchers. I discussed this with two PhD students from my group who told me that they don't feel comfortable asking questions about topics they may not be familiar with. (6/7)

Show thread

4 - Of course the last one is an implicit rule arising from
the fact that ARCHES is a small "private" conference (only members of the organizing teams can attend). (5/7)

Show thread

3 - friendly atmosphere and respectful scientific interactions so that junior attendees don't feel afraid to present their work or to ask question. (4/7)

Show thread

2 - senior members do not present their own work but an overview of the work of their research team members; (3/7)

Show thread

1 - minimal registration fees, we don't need fancy dinners or luxurious venues; (2/7)

Show thread

I have just returned from ARCHES, a small European conference in audiology. The organizing comittee (which my team is part of) follows a set of self-imposed ruled which I think should be a model for other international scientific meetings: (1/7)

@kinozhao @philosophy @academicchatter

FWIW, I'm not a professor but as a recent undergrad, I can confirm previous research saying notes don't help study anyway, self-testing does. So withholding them won't help them make better notes.

Additionally, if you're withholding slides but then examining their understanding of the content using specific, definable criteria for what is correct/incorrect, then you better be very explicit about what you're actually teaching in your lectures.

what if we all agreed to use each other's target stimuli as fillers in syntax experiments
#linguistics #psycholinguistics #experimentalLinguistics

A new paper building on Wong et al (2022) again shows that the misuse of Bayes factors is prevalent, and as far as I can see *massively* higher than the misuse of frequentist p-values: psyarxiv.com/du3fc/. Really appreciate these proponents of Bayes factors taking such a critical look at how Bayes factors are used. I think Bayes factors will soon join p-rep and 'the new statistics' as a failed alternative to p-values. 1/2

Vous faites de l'#IA #nlp ? de l'informatique ? Vous allez avoir besoin de ce bingo #éthique (réalisé avec @emilymbender) :

RT @pskatz@twitter.com

There seems to be an increasing use of the justification for a paper being "little is known". Here is the appearance of this phrase in PubMed. It looks like we know less as time progresses. Little is known about why this trend has developed.

🐦🔗: twitter.com/pskatz/status/1592

Ceci souligne deux choses :
- d'une part que les performances des aides auditives sont aujourd'hui globalement plafonnées techniquement, il n'y a pas d'innovation cruciale récente qui ferait qu'un modèle ou une marque se détache du marché (même si individuellement on peut avoir une préférence pour un modèle particulier)
- d'autre part que grâce à la réforme 100% santé la situation actuelle est telle que la grande majorité de la population peut accéder gratuitement à des aides auditives à peu près équivalentes aux meilleurs modèles, ce qui est une excellente nouvelle.

Show thread

Par ailleurs, cette conclusion concorde avec la littérature scientifique sur le sujet.

Show thread

Les résultats de la dernière enquête Que Choisir sur les audioprothèses confirment ceux de la précédente : les aides auditives bon marché (entièrement prises en charges pour les personnes bénéficiant du "100% santé") sont globalement équivalentes aux appareils plus haut de gamme. quechoisir.org/enquete-auditio
(Attention le test est réalisé sur avec un réglage par défaut donc sans le réglage individualisé réalisé par l'audioprothésiste)

KataGo, one of the best #Go Playing #AI can be defeated by amateurs, using a simple trick... which is easily countered at amateur level.
arxiv.org/abs/2211.00241

"Even professional-level AI systems may harbor surprising failure modes". Imagine similar failures in critical systems (self-driving cars). There is no better summary of the state of the art in #MachineLearning.

At long last our paper

"Neural dynamics of phoneme sequences reveal position-invariant code for content & order" is out!

- paper: nature.com/articles/s41467-022
- data: osf.io/ag3kj/
- press: nyu.edu/about/news-publication

By our NYU dream team @GwilliamsL @AlecMarantz & @DavidPoeppel

Show older
Qoto Mastodon

QOTO: Question Others to Teach Ourselves
An inclusive, Academic Freedom, instance
All cultures welcome.
Hate speech and harassment strictly forbidden.