Public notes (Page 7)

Clean Code: Separating abstractions from concretions

Uncle Bob reminds that source code dependencies should cross the boundaries between the abstract and the concrete pointing from the concrete side towards the abstract side. Concrete -> Abstract As in: Application and Domain Objects ^ | DB Interface Layer | v Database Source: Clean Code

Clean Code

Clean Code: Classes and Data Structures

From SOLID principles, your code should be open to extension, and closed to modification. As Uncle Bob points out, classes protect us from new types, while data structures protect us from new functions. This means that when you suspect new functions may be required

Deep Learning

Jeremy Howard: If your training loss is higher than your validation loss, it definitely means that you are underfitting.

"You want your training loss to be lower than your validation loss" - said Jeremy. When you are underfitting, you can: train for longer, train the last bit at a lower learning rate. If you are still underfitting, then you are going

Information Sciences

Jeremy Howard: Using smaller images in training first works great, but is still largely unknown.

Jeremy called that idea "progressive resizing". Knowing that starting from smaller images turns out to accelerate learning and results in better generalizations, an even more precise term for this technique would seem to be "progressive upsizing". One of his findings

Information Sciences

Jeremy Howard: In transfer learning you start with a model trained to do something, then fine-tune it.

When you have a model trained to recognize images of size 128x128, you don't need to train another model from scratch to be able to recognize images of size 256x256. What you can do instead, is: retrain the prevoius model for the new task.

Information Sciences

Frank Nielsen: Information sciences seek methods to distill information from data to models.

In a recent study published in a scientific journal named Entropy, Frank Nielsen proposed the term "Information Sciences" to describe the common goal of several modern efforts to turn data into models: In short, information sciences seek methods to distill information from

Deep Learning

Jeremy Howard: Once the data is there, there is very little to do afterwards.

What you can learn from Jeremy Howard from Fast.AI is that preparing the data is the crucial step in preparing a state-of-the-art Deep Learning model. For classifying cat and dog images, it is good to rotate some of the pictures; flip some of

Software Engineering

Windows 10: Kolory w konsoli

Skonfiguruj kolorowanie składni w konsoli pod Windows - prawie jak w Linuksie!

offer

Code review offer

I once submitted my code for review. At the time I had 3,5 years of professional programming experience. When I got the answer, I was amazed with the amount of insight I got from just that single submission. My reviewer was a person

Functional programming

Bartosz Milewski: You can sometimes replace functions with data.

Nature

Emmy Noether: Every conservation law corresponds to some kind of symmetry.

To illustrate the whole problem still more clearly, imagine that we were talking to a Martian, or someone very far away, by telephone. We are not allowed to send him any actual samples to inspect; for instance, if we could send light, we could

Nature

Bartosz Milewski: Jesteśmy w stanie tylko to poznać - co się dekomponuje.

QED: Part 1

Richard Feynman said about QED that students do not understand this theory, and that is because the professors do not understand it either! NATURE Venus sometimes is a morning star, and sometimes - an evening star! INVENTION Feynman: Mayans had hundreds of thousands of

Physics

QED

The crown jewel of contemporary physics

Haskell

Haskell and cryptocurrencies

Many blog posts about Haskell often discuss the latest advances in our compiler, research in type systems and clever new ideas that make the Haskell language such a fun and inspiring tool. However, if you peel back the curtain on a lot of what

nature

Po prostu przeczytaj: Jak wynaleźć wszystko

Nature

Hawking Radiation

When a black hole swallows a portion of matter, that matter ceases to have any energy. Its energy falls to absolute zero. This means that for that portion of matter,

Nature

5 wciąż nie rozwiązanych problemów fizyki

Pod koniec XIX wieku fizycy mieli poczucie, że odkryli już wszystko. Zaskoczyło ich odkrycie Heisenberga, a to był dopiero początek. Również dziś liczne problemy fizyki pozostają bez zadowalającej odpowiedzi. Tu znajdziesz kilka ich przykładów.

Quantum fluctuations

Quantum fluctuations, as they exist, stem directly from the principle of uncertainty. Until 1997, there was no sure evidence behind the theory that quantum fluctuations really take place. Then, a team of scientist conducted the experiment with enough accuracy for it to be valid,

Human-friendly Haskell

For months I have been wondering how to abstract a "class" meaning "a set of reusable functions" in Haskell. Finally I found a sample solution: class

Rozmowa kwalifikacyjna - JavaScript + schematy

JavaScript Operator == a ===, IIFE Operacje asynchroniczne Hoisting AngularJS Serwisy

Mity o miłości

Oto notatki z nagrania z konferencji z ks. Markiem Dziewieckim w ramach cyklu Sympatia Miłość Małżeństwo "Miłość w wielkim mieście" w Szkole Głównej Handlowej 4.05.2017. Mit 1: "Miłość to współżycie" Miłość nie jest tym samym co pożądanie czy

JavaScript: Living to see a Promise resolved after a second

Promises in JavaScript are an attempt to escape the need to define a callback on top of a mountain of other callbacks. Too often we find ourselves defining a callback which is a callback to a callback which is called by another callback. However,

Nature

Genomes and results

From all we know, the genome influences a large portion of the organism's features. We also know that the human genome can be encoded using around 750 MB of data.

Public notes

Elegance is not optional.

Clean Code: Separating abstractions from concretions

Clean Code: Classes and Data Structures

Jeremy Howard: If your training loss is higher than your validation loss, it definitely means that you are underfitting.

Jeremy Howard: Using smaller images in training first works great, but is still largely unknown.

Jeremy Howard: In transfer learning you start with a model trained to do something, then fine-tune it.

Frank Nielsen: Information sciences seek methods to distill information from data to models.

Jeremy Howard: Once the data is there, there is very little to do afterwards.

Windows 10: Kolory w konsoli

Code review offer

Bartosz Milewski: You can sometimes replace functions with data.

Emmy Noether: Every conservation law corresponds to some kind of symmetry.

Bartosz Milewski: Jesteśmy w stanie tylko to poznać - co się dekomponuje.

QED: Part 1

QED

Haskell and cryptocurrencies

Po prostu przeczytaj: Jak wynaleźć wszystko

Hawking Radiation

5 wciąż nie rozwiązanych problemów fizyki

Table of contents

Quantum fluctuations

Human-friendly Haskell

Rozmowa kwalifikacyjna - JavaScript + schematy

Mity o miłości

JavaScript: Living to see a Promise resolved after a second

Genomes and results