Toluwani Aremu

Psalms 24:1 | PhD Candidate @MBZUAI | Visiting @ASTARsg | Building Quant stuff @theKoraHQ | Safe and Trustworthy AI | 👇🏽👇🏽 for more

[Toluwani Aremu @tolusophy]

@ElijahSchaffer See how he left the place without stabbing anyone? Exactly!

[Toluwani Aremu @tolusophy]

@cryptojourneyrs At the end of the day, everyone has to hold themselves accountable, and that includes us black people, especially those crying wolf about the outcome. We are all just humans first, before race comes into play.

[Toluwani Aremu @tolusophy]

@JeffDean I think they wanted more visibility for the world cup. They've gotten that now.. I wouldn't be surprised to see him officiating some of these games. 😂

[Toluwani Aremu @tolusophy]

@SuJinyan6 Was science ever truly open? Loopholes (IP, secrecy, competition) existed for centuries. What changed is visibility + stakes imo. The real risk is pretending it was perfect before.. And the conclusion is that transparency will continue to blur over time.😬😬😬

[Toluwani Aremu @tolusophy]

@itsManghwani @SuJinyan6 That's Human in the Loop😂

[Toluwani Aremu @tolusophy]

@kimmonismus I don't see a contradiction here... AI is disruptive, but the current state of AI can hardly replace people we have behind core infrastructure... But it is coming anyway

[Toluwani Aremu @tolusophy]

@SpartacusAFC @anishmoonka What I'm saying is, there's no excuse to not winning a continental competition even when you've dominated your local one. It is a game of "it is meant to be"... Anything can happen

[Toluwani Aremu @tolusophy]

@petrogustavo Don't be scared people, he is just learning English...

[Toluwani Aremu @tolusophy]

@SpartacusAFC @anishmoonka Real Madrid knocked out man city, or is the La Liga a farmer's league too? Probably a farmer's league with machineries lool... My argument still holds

[Toluwani Aremu @tolusophy]

@Tomiwakol @anishmoonka That's why it is called a "saying."

Arsenal almost won the game against PSG if not for the penalty. Thank God 😅...but you see, may the best team win means we don't care who played the better football, we award who executes their plan better! Vox Populi Vox Dei...

[Toluwani Aremu @tolusophy]

@haider1 Is it clever? Sure
Is it conscious? No

No worries then...

[Toluwani Aremu @tolusophy]

@kimmonismus Full respect!

[Toluwani Aremu @tolusophy]

Africa's poverty is not just about colonialism. Yes, colonialism definitely took advantage of resources, but asides the fact that we did enable this exploitation, we also helped keep things that way through our own choices after independence. And look at Ethiopia. They were never colonized, yet they're still poor because of their mismanagement and conflicts.

On the other hand, Botswana did very well with good governance and how they managed their diamonds, becoming a great success story for Africa. South Korea really bounced back from the war, went through some reforms, and became one of the OECD powerhouse countries. Their currency is similar in strength to the naira but their economy is so strong they can manage to pay a good minimum wage.

For us in Nigeria, sub-Saharan Africa at large, corruption is a huge problem, averaging about 32% CPI. THAT'S WAY HIGHER THAN THE GLOBAL AVERAGE. On top of that, institutions are pretty weak. Blame our woes less on history and more on lack of accountability and dishonest leadership! Look at the ones we have there right now! Look at the ones that are trying to run... We should also look at ourselves. We should stop celebrating victimhood and demand for transparent leadership, not promise givers and breakers like those we've had in the past, those we currently have, and those that are pretending to care enough to run.

[Toluwani Aremu @tolusophy]

@masuzafi *female president*
Should there be a distinction?

[Toluwani Aremu @tolusophy]

@theCryptoglyph @idris_lfc1 Whatever makes you sleep well at night sir 🙏🏿

[Toluwani Aremu @tolusophy]

If you read that point well, you would see that I never said Salah came as pure goal machine. However, this doesn't take away the fact that Salah has 45 goal contributions in 61 matches for Roma (https://t.co/8y9Ughm4ah).

Back to my point, I stated "9 season as a pure goal machine," not "he came as pure goal machine."

Shey you get?

[Toluwani Aremu @tolusophy]

@mfc_mrpage Argentina

[Toluwani Aremu @tolusophy]

Elegance is a feature, not a bug.

[Toluwani Aremu @tolusophy]

Sometimes I wonder why reviewers reward very complex systems proposed to solve a problem that seems rather simple though significant. I don't believe in a total reinvention or revamp of current systems just for this.

For example, in watermarking, one very significant security issue is forgery (or spoofing). A lot of watermarks commonly deployed in practice can be forged, and the most common form of method attackers use comes from distilling a spoofer on watermarked outputs from the provider model.

In our work (https://t.co/h7VTDxvKqE), we propose a simple yet very effective solution to this. Basically, we tell the providers, do not change your watermarking method. Instead, increase your number of keys, and let your system sample from a key during each generation. Then, alter your detection method a bit to find out if a watermarked response contains more than one key. If it does, definitely there's something fishy about it

We didn't reinvent the whole system, infact, per sample robustness is guaranteed as we inherit it from the base watermarking method. We are just improving the forgery resistance of that watermarking method to make it a more complete system (robustness + security). Our empirical results are so impressive that they smash our theoretical expectations.

Yet, I see over-complicated, painful-to-watch methods get accepted while simple, deployable fixes get scrutinized. A friend advised me to complicate my writing to "feign rigor." I can't do it. I want a layperson to understand my work. I am a big fan of simple, effective, and efficient solutions.

Anyway, reviewers should normalize adopting and appreciating methods that do not seek to fully reinvent the wheel to solve an important problem.

Is the problem significant? Yes
Does the method solve it? Yes
Is the method simple to implement? Yes
Is it deployment-ready for existing system? Yes
Is the application novel? Yes

Accept the paper goddamnit 😂! Thank God for preprint servers!