AlexTCH
313 subscribers
77 photos
4 videos
2 files
906 links
Что-то про программирование, что-то про Computer Science и Data Science, и немного кофе. Ну и всякая чушь вместо Твиттера. :)
Download Telegram
Наконец-то по-настоящему полезный ИИ! 😁
Forwarded from Brenoritvrezorkre
elicit.org, некий рисёрч ассистент нейросеть
https://haslab.github.io/formal-software-design/index.html
"Formal Software Design with Alloy 6"

Includes a short section on teaching with Alloy4Fun and primers on relational and temporal logics for the reference.
But didn't read yet.
https://www.youtube.com/watch?v=85fiogkCyLY
Fundamental Computing group of the University of Groningen пиарится и пытается объяснить "простым смертным" чем они занимаются (спойлер: формальной верификацией программ).
Написал "poerational semantics". В принципе, так оно и есть... 😂
Новогодняя отъебутика от нашего e-бутика!
Wham! письмо!
https://youtu.be/lAU5hx_3xRc

A really nice lecture from Xavier Leroy giving an overview of critical software development and pretty much all of existing approaches to verification. With some entertaining examples from avionics.

And a gem from a question: "Fixing bugs is not research". I'm not sure much changed since 2016...
👍1
Apparently the computational complexity of gradient descent belongs to some obscure class. 😁
https://arxiv.org/abs/2011.01929
Оказывается, для "плоских" (один скрытый слой) нейронок с ReLU (нелинейной, между прочим) активацией, градиентный спуск из случайной точки (случайные начальные веса) доказанно находит глобальный минимум (при использовании квадратичной функции потерь): https://arxiv.org/abs/1810.02054
👍1
https://www.cs.utexas.edu/users/wcook/Drafts/2009/essay.pdf
"On Understanding Data Abstraction, Revisited" by William R. Cook, 2009

A fascinating essay drawing clear lines between Data Abstraction, Type Abstraction, Abstract Data Types and Objects.

For extra fun controversial rhetoric the author argues that in its essence OOP is just Higher-Order Functional Programming and even states that
One conclusion you could draw from this analysis is that the untyped λ-calculus was the first object-oriented language.

😁

But I see his point. For me personally that reminds https://www.elegantobjects.org/ which in pursue of "true OOP" embraces many FP mechanisms and techniques. Moreover on a couple of occasions the author mentions that processes could model objects as well which reminds another famous saying that Erlang is the most OO language. 😊

Fun controversy aside the author presents pretty deep analysis explicitly based on seminal works of Luca Cardelli and Peter Wegner, as well as more resent "Types and Programming Languages" presentation. Essay touches on ADT and OO relationships to extendability, flexibility, optimisation, Expression Problem (with excellent bibliography going all the way back to Reynolds) and verification. Significant chunk of the essay is devoted to discussion of actual programming languages' (Java, Smalltalk and Haskell) relationship with both ADTs and Objects, and features that support or hamper their implementation and sound use.

At any rate that was an enriching read from both theoretical and practical perspectives, a bit too brief but therefore even more thought-provoking. 😊
Поскольку "новое быдло" звучит токсично и неполиткорректно, мы говорим "пост-интеллигенция".
https://rudalle.ru/check_image/d7b2e54ffac845e9a5d3bba8cc1961e9

Диаграмма классов типичного Java-приложения. 😁