Julien Girard @rhapsodos@scholar.social

Btw here are the speakers' book: https://www.combinedacademic.co.uk/9781503609747/the-costs-of-connection/

Sessions endend just now, and it was very interesting. It gave me a brand new insight on how to understand the new age of data collection as a continuation of colonial dispossession, I will dig this up

I will particularly be attending today's session, about Digital Surplus and Data Colonialism.

One of the researcher from my team is organizing a workshop on AI consequences on society: how those technologies can amplify our racial and sexist biases, or make jobs even precarious? Those are some of the questions this seminar aims to answer:
https://fondazionefeltrinelli.it/eventi/unboxing-ai-understanding-artificial-intelligence/

Inscription is free but mandatory, feel free to join. 15h-19h from today to saturday

I think there is this issue with PhD studies where no one, except the student, can assess the quality of the work. If you have either a low self-esteem or high expectations, things can go spiraling down real quick. The student is the only one deciding on a day-to-day basis when they should stop working or what is to be considered "a good working day". There are no objective, outside metrics that helps to assert that, because almost every research activity is peer-based.

I have an intern since two month or so, and although it is incredibly rewarding to see him grow and become more and more capable, I still fear that he does not learn "enough". Of course there are goals for his internship (we will soon submit a paper), but I don't know what he should retain from this internship, how should a "good researcher" behave.

Heya. The materials for my tutorial at PFIA 2020, "Formal Verification of Deep Neural Networks", are available here:
https://gitlab.inria.fr/jgirard/pfia-2020

The tutorial will take place monday morning at 10 a.m. CET

You can subscribe to the webinar here: https://webikeo.fr/webinar/tutoriel-1-verification-formelle-de-reseaux-de-neurones-profonds-theorie-et-pratique

A paper, named " A Deep Neural Network Model to Predict Criminality Using Image Processing", will be published in a Springer venue. You might find the title unsettling. Well, guess what, it is. Using data produced by the criminal system will not "predict crime", it will predict who the police arrests. Which are two really different things.

More details there:
https://medium.com/@CoalitionForCriticalTechnology/abolish-the-techtoprisonpipeline-9b5b14366b16

I'll be giving a tutorial on formal verification of deep neural networks at PFIA 2020. All material (slides, code) will be distributed freely.
http://pfia2020.fr/tutoriels/programmetutoriels/

A quite pessimistic point of view of french security and cryptograpy experts on a COVID-19 tracking app (in French only):
https://risques-tracage.fr/

Fifteen simple abuse case scenario, a state of the art of the possible form such an app coud have... an interesting reading

An interesting insight on how Facebook moderation build an hierarchy of "protected categories" of people:
https://www.theguardian.com/news/gallery/2017/may/24/hate-speech-and-anti-migrant-posts-facebooks-rules

PJ Reddie, one of the most wholesome and cool scientists of this world, quitted research in computer vision due to ethical concerns:
https://nitter.net/pjreddie/status/1230524770350817280#m

A cool conference on the deep learning hype, at 36c3:
https://media.ccc.de/v/36c3-11006-der_deep_learning_hype

Good questions here:
* is deep learning a sustainable technology in our resource-limited world?
* are those deep learning systems useful for anyone else than big companies and a restricted set of academics?
* What about the anonymous workforce used to label or mimic deep learning datasets?
* how the heck can we continue to publish to top-tier conferences with 10% of results that are not reproductibles

A petition from some Machine Learning researchers to encourage major conferences to offer a remote participation option:
https://www.change.org/p/organizers-of-data-science-and-machine-learning-conferences-neurips-icml-aistats-iclr-uai-allow-remote-paper-poster-presentations-at-conferences?recruiter=144678980&recruited_by_id=51a298c0-2d0e-11e4-843e-ad2ec88f7f02&utm_source=share_petition&utm_medium=copylink&utm_campaign=petition_dashboard

If I were a scholar on economical history and philosophy, I would say that it's just capitalism trying to devour another field. That after reaping everything it could from our diseases, our culture, our food, our emotions, it goes for our knowledge and spirit.

But I am none of those. I am just a poor PhD student raging against this state of affair, and hoping that things will change.