Fernando Rios Cruz

Consultor CONTPAQ i

[Math Pulse @themathpulse]

The Beauty of Math

🎥 Alan Becker

[Fernando Rios Cruz @frioscruz]

@RoyCampos Que triste papel alguna vez te consideré un encuestador creíble

[Fernando Rios Cruz @frioscruz]

@RoyCampos En el pasado te escuchaba con atención antes de Morena, hoy cuando te escucho cambio de estación

[Fernando Rios Cruz @frioscruz]

@RoyCampos Roy deja de darle cuerda a Morena

[Fernando Rios Cruz @frioscruz]

@StratCons Yo tengo que ir cada mes a una consulta para que me surtan la tamsulusina para la próstata ocupando un tiempo de consulta para un paciente con problemas mayores😂

[Alex Zajac (🍔,🧠) @itsalexzajac]

An engineer built an interactive book on data structures and algorithms of 680 pages:

[Tivadar Danka @TivadarDanka]

In machine learning, we use the dot product every day.

However, its definition is far from revealing. For instance, what does it have to do with similarity?

There is a beautiful geometric explanation behind.

[Girişimci Hisler @girisimcihisler]

STANFORD 200 BİN DOLARLIK YAPAY ZEKA MÜFREDATINI TAMAMEN ÜCRETSİZ ERİŞİME AÇTI!

Mülakat yok, başvuru bedeli yok, sadece belirli bir zümreye ayrılmış kontenjanlar yok. Stanford Üniversitesi, en gelişmiş yapay zeka modellerini inşa eden mühendisleri eğittiği orijinal ders kayıtlarının tamamını YouTube üzerinden açık kaynak haline getirdi.

Pazarlama amaçlı hazırlanmış, "herkes için yapay zeka" tarzı yüzeysel bir başlangıç kursundan bahsetmiyorum. Bu doğrudan işin mutfağı. Bugün OpenAI veya DeepMind gibi şirketlerde devrim yaratan uzmanları şekillendiren o ağır müfredatın ta kendisi.

İşte yeni yayınlanan üst düzey teknik eğitim serileri:

Derin Öğrenme (CS230)
→ https://t.co/vYlS1r7PTD

Transformer Mimarileri ve Büyük Dil Modelleri (CME295)
→ https://t.co/vYlS1r7PTD

Sıfırdan Dil Modeli Geliştirme (CS336)
→ https://t.co/kunhALlheS

İnsan Geri Bildirimiyle Makine Öğrenimi - RLHF (CS329H)
→ https://t.co/UmTvocENs7

Bilgisayarlı Görme ve Görsel Algılama (CS231N)
→ https://t.co/xhgUeHOHih

LLM Değerlendirme ve Ölçeklendirme Süreçleri
→ https://t.co/nwD9JvqtU4

Sektörde hepimizin yüzleşmesi gereken çok net bir gerçek var: Diplomalar artık eskisi kadar nadir veya tek başına yeterli değil. Asıl nadir olan şey, bu teknik bilgiyi alıp hızla gerçek bir ürüne dönüştürebilme becerisi.

Dünyanın en prestijli eğitim kurumları oyunun kurallarının değiştiğini çok iyi biliyor. Tüm bilgi birikimlerini herkesin erişimine açmalarının asıl sebebi de bu rekabet.

Bu devasa arşivi kaybetmemek için mutlaka kaydedin.

[Prof. Fabrizio Bucella @FabrizioBucella]

La conjecture de Riemann concerne les nombres premiers. Tous les zéros non triviaux de la fonction zêta de Riemann ζ(s) ont pour partie réelle ½. Autrement dit, ils sont tous alignés sur une seule droite verticale du plan complexe, la droite critique.

L'astuce. Si vous obtenez les zéros de la fonction zêta de Riemann, alors vous avez encodé tous les nombres premiers.

Le prix. C'est l'un des sept Problèmes du Millénaire fixés par le Clay Mathematics Institute en 2000. Six prix restent ouverts (Poincaré ayant été résolu par Perelman en 2003, qui a refusé le prix).

La spirale. C’est le portrait de la fonction zêta quand on la promène le long de la fameuse droite critique Re(s) = ½. À chaque valeur de t, on obtient un nombre complexe ζ(½ + it), qu'on place dans le plan. La courbe que vous voyez, c'est l'enchaînement de toutes ces valeurs.

Chaque fois que la spirale traverse l'origine, c'est qu’à ce moment précis ζ vaut zéro (un zéro non trivial).

Le premier arrive à t ≈ 14,13 ; puis 21,02 ; 25,01 ; 30,42... Sur 11 secondes de tracé, vous en voyez environ vingt.

La conjecture de Riemann affirme que toutes les valeurs de s qui annulent ζ se trouvent sur cette droite et pas ailleurs dans le plan complexe.

Personne ne sait le prouver depuis 1859.

Salukes

✅​ Simulateur disponible sur le Linktree. La science peut tout.

#science #physique #profbucella #lasciencepeuttout

[Suneeta Vardarajan @Ms_Gravitas]

A note on the Poincare conjecture, and how Perelman solved it: Poincare's conjecture was about three dimensional spaces without boundary. 1/n

[Fernando Rios Cruz @frioscruz]

La bu

[Michał Podlewski @trajektoriePL]

The Polish theoretical physicist who proved that you can recreate all mathematical functions from just this one operation is Andrzej Odrzywołek from Jagiellonian University @JagiellonskiUni. Truly a remarkable piece of work! Congrats @AndrzOdrz
https://t.co/vnWJB4JCmf

[Mike @mike_mates]

Parace que han demostrado que una función de dos variables puede generar a todas las demás funciones "elementales".

Osea que con esta calculadora de dos botones podemos calcular cualquier función usual.

Es como la puerta NAND de las funciones elementales.

[NASA Ames @NASAAmes]

Apollo then, Artemis now.

A set of Earthset images captured by the Artemis II crew during their lunar flyby on April 6, as well as Earthrise photos taken during the Apollo 11,12, and 17 missions.

Image descriptions available in the comments below.

[santi @santtiagom_]

https://t.co/z2mzS1CuY8

[Rohit Kumar Tiwari @_rohit_tiwari_]

This playlist has 26 deep learning lectures.
One of the best free resources online.
Go from beginner to confident in just 5 hours. https://t.co/KjaS7jAt0S

Foundations of Deep Learning
> What deep learning is
> The math basics you need, including linear algebra and calculus
> How a single neuron works as a computing unit

Training Mechanics
> How a neuron is trained using gradient descent
> Why the data analysis pipeline matters
> How out of sample validation helps check model reliability

Network Architectures
> Feed Forward Neural Networks explained
> Backpropagation and why it lets multi layer networks learn from errors

Optimization for Classification
> Activation functions like Softmax
> Loss functions like Categorical Cross-entropy
> How these handle complex classification problems

Efficiency and Stability
> Making networks faster with vectorization
> How to spot and fix vanishing or exploding gradient issues

Generalization and Regularization
> Ways to avoid overfitting
> How to make sure models work well on new data, not just training data

Computer Vision and Transfer Learning
> Convolutional layers for image tasks
> Using transfer learning and data augmentation
> Training large models with limited data

Advanced Network Features
> Residual Networks explained
> Why skip connections help build very deep models

Natural Language Processing
> How machines understand text
> Word embeddings
> Recurrent Neural Networks and LSTMs
> Transformers and self attention

Generative and Specialized Models
> Auto Encoders for data compression
> Generative Adversarial Networks for creating new data
> The ideas behind AlphaGo and reinforcement learning

[Fernando Rios Cruz @frioscruz]

@malvarezvargas Lo barato sale muy caro

[Lore Cárdenas. @LoreCardenasX]

Tengo 47 años
El estrés me atrapó durante años, así que me obsesioné con la neurociencia.

Después de más de 10 000 horas investigando a personas de alto rendimiento, directores ejecutivos y genios, mi mente ahora es invulnerable. He ayudado a más de 493 personas a lograr lo mismo.

Estos son mis 5 trucos neurológicos favoritos que te cambiarán la vida:

[Fernando Rios Cruz @frioscruz]

@ecommartinez "4.6"

[Fernando Rios Cruz @frioscruz]

@ecommartinez 4.6