Heya. The materials for my tutorial at PFIA 2020, "Formal Verification of Deep Neural Networks", are available here:

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

You can subscribe to the webinar here: webikeo.fr/webinar/tutoriel-1-

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:

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

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

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

it's gross and creepy to see patterns in sites like Netflix where I'll watch a few films in Spanish or watch a couple films with Black leads, or i'll watch a few LGBTQ romances, and then all of a sudden the whole site changes. film covers are changed to display BIPOC folx more often, i somehow unlock whole new categories of films, etc.

then i watch a couple of blockbusters one weekend and the site reverts to a homogenized safety bubble of middle class hetero white society

it's scary to think how many sites create these safe spaces for privileged people to erase everyone not like them, and no one even notices

Coucou Mastodon !
Jeudi 5 mars, je suis en grève pour l'action "La recherche s'arrête"
J'aimerais toucher les élèves à qui j'enseigne. J'ai l'impression qu'on les exclut souvent de nos luttes; ils sont de plus assez peu politisés (l'informatique, tout ça)
J'aimerais détourner les deux heures que j'ai pour discuter avec eux de la réforme, de politique en général...
Est-ce que vous avez des conseils de comment procéder ? Des supports ?
Boosts appréciés !!

Hello! I'm a researcher at Harvard's Shorenstein Center working on disinformation, censorship, and media manipulation in general. Glad this place exists :)

PJ Reddie, one of the most wholesome and cool scientists of this world, quitted research in computer vision due to ethical concerns:

A relatively large group of French scientists propose to add a fictional Alex to all our publications in all domains. This would make the Alex an extremely prolific scientific publisher:


I think (and the people behind this idea do as well) that this initiative may be interesting to researchers all over the world. Unfortunately, the page they provide is in French. Maybe they will set up another one in English later.


A cool conference on the deep learning hype, at 36c3:

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

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.

Show thread

What is happening in research (and other public sectors) is a colonization. All are colonized by a foul mindset: looking for material value, immediate profit and systematic exploitation.

Well guess what: trying to quantify how valuable is a knowledge, how much money an idea may give... it simply does not work, and I think it is wrong to think this.

Favouring ideas that are economically rewarding will simply enslave research to economy, preventing to look for new perspectives.

Show thread

All recent reforms are destroying what is making academia unique, engaging, rewarding and fun. I think that researching knowledge for a society is a wonderful job. Teaching, when voluntary, is a genuine gift. It empowers people, helping them to reassert their situation, giving them tools and strengths to face the future.

Show thread

Neg about academia work condition 

I fully support their action. As a young PhD student, working in french academia is a bleak perspective. Competition is insanely strong, salaries are quite shitty when compared to engineers, analysts, or people with a technical background and about the same years of study.
Psychological hazards in research, which are already quite huge, would certainly increase when removing material and intellectual liberty for researchers.

Show thread

Hop, some time for french politics (don't flee yet, this might be interesting).

Over the recent years in France, there is a global trend to privatize a LOT of previously public-owned services. Transportation, energy production... and public research funded by the state is no exception.

Some academics wrote a statement about this. They are planning a team application to one of the french "research quality assessment group" rogueesr.fr/hceres-en/

Voici un texte de RogueESR que j'approuve sans réserve. Aux académiques: n'hésitez pas à signer. Aux non académiques: je sais que je me répète, mais regardez ce qui est dit, car comme iels le disent "Sans recherche autonome, nous n’avons pas d’avenir."

My paper on formal verification of deep neural networks using simulators was accepted at the European Conference on Artificial Intelligence!

CAMUS: A Framework to Build Formal Specifications for Deep Perception Systems Using Simulators

I will be happy to discuss about it!

(I am building up my hype for the congress I will speak)

Philip Wadler is speaking right now at College de France, about using a proof assistant during teaching.

The talk will be available freely on the website of College de france


Show more
Scholar Social

Scholar Social is a microblogging platform for researchers, grad students, librarians, archivists, undergrads, academically inclined high schoolers, educators of all levels, journal editors, research assistants, professors, administrators—anyone involved in academia who is willing to engage with others respectfully.