A surprising situation.

A machine learning model predicting 3 classes ends up with these results:

• Class A: 90% accuracy
• Class B: 80% accuracy
• Class C: 70% accuracy

Class C seems to be the worst-performing.

But you know already that I'll say it isn't. ↓

Here is the correct answer: We don't know yet which class is the worst.

This is important to understand, so keep reading.

Many people make this same mistake every day and try to fix things that ain't broken.

We are missing something, so let's go and get it.

How about if we ask a friend to look at the dataset manually?

Let's say these are the results of our friend doing the job:

• Class A: 95% accuracy
• Class B: 100% accuracy
• Class C: 71% accuracy

Do you see what just happened?

The model is actually really accurate when predicting Class C! 70% versus 71% when done by a human.

Class B, however, is a disaster: 80% versus 100% when done manually.

Class B seems to be the worse performing class!

The work that our friend did is called a "baseline."

Without having it, it's hard to determine how good are the results of a model.

I've seen tons of money down a drain just because teams didn't create a baseline beforehand.

But now, that won't happen to you 😉.

5 JavaScript Animation Libraries Every Web Developer Should Know

A Thread 🧵👇

1. Swiper

Swiper is the most modern free mobile touch slider with hardware accelerated transitions and amazing native behavior. It is intended to be used in mobile websites, mobile web apps, and mobile native/hybrid

2. GreenSock

The GreenSock Animation Platform (GSAP for short) is a powerful JavaScript library that enables front-end developers and designers to create robust timeline based

3. AOS Library

AOS - Animate on scroll library. It has no other dependencies, which helps to keep your website's performance under

4. Three.js

Three.js is a cross-browser JavaScript library and application programming interface used to create and display animated 3D computer graphics in a web browser using

Are you planning to learn CSS for adding some cool styling in your website?

In this thread, I'm covering all the basics and proper plan you need to know in order to getting started with CSS.

🧵 👇🏻

We need to know the importance of CSS first. Why CSS?

Well I guess CSS is the only language that add some visuals in website. Without CSS, Without it, websites would still be plain text on white backgrounds.

{ 02 / 17 }

Let's compare a website with and without CSS

Image 1: With CSS
Image 2: Without CSS

{ 03 / 17 }

I this should be enough to explain you why CSS is so important. So lets start

You can gain some decent amount of CSS knowledge in 30 days. And after that you will definitely able to create some good looking website. Let's see how 👇🏻

{ 04 / 17 }

First things first, you need to learn basic stuff

- How to link your CSS files with HTML file
- How to select elements so that you can add styling

{ 05 / 17 }

5 ways to get your first client

🧵👇

1. Facebook Groups

There are 2 ways to go about this:

1.1. Join existing Facebook groups of your target clients.

For this one, you need to add value by commenting, answering questions & giving suggestions.

Do not sell your services right away, give value first, then pitch.

1.2. Create your own Facebook group and add your target clients as members.

Some tips for your group:

- Interact with your members
- Add valuable articles and videos
- Pitch your services once a month

2. LinkedIn Outreach

- Use InMail Messages
- Connect with 20-30 profiles of your ideal client every day

While doing this, post valuable tips (for your niche) and after a while, reach out to them to offer your services.

(LinkHelp .io is a great tool to automate this)

3. Freelancing Platforms

How I did it:

- I charged $5 for a website analysis report
- I upsold to a $100 wireframe
- I upsold to a $1k website
- I upsold to a $1k marketing retainer

Don't overlook starting with "small" services.

→ Fiverr, UpWork, Hubstaff Talent

In a new paper, our team tackles a fundamental AI problem: how can we simultaneously parse the world into objects and properties, while simultaneously inducing the rules explaining how objects change over time: https://t.co/f7mZMtwu6z (1/)

Work by @LittleBimble with @pfau, @pushmeet, Matko Bosnjak, Lars Buesing, Kevin Ellis, and Marek Sergot. (2/)

This system combines the Apperception Engine with a binary neural network to learn a provably 100% accurate model of non-trivial environments (e.g. Sokoban) from noisy raw pixel data. (3/)

The ability to learn the world dynamics sample efficiently is a key component of intelligence, according to @MelMitchell1, @GaryMarcus, @fchollet, @mpshanahan, among many others. (4/4)

7 GitHub Repositories (Self Taught) Developers Must Follow

A Thread 🧵

1. Free Programming Books

This repository contains tons of free programming books. It has grown to become one of GitHub's most popular repositories, with 160,000+ stars, 6000+ commits, 1600+ contributors, and 39,000+

2. Coding Interview University

This is a multi-month study plan for going from web developer (self-taught, no CS degree) to software engineer for a large

3. Developer Roadmap

Here you will find a set of charts demonstrating the paths that you can take and the technologies that you would want to adopt in order to become a frontend, backend or a

4.

Correlation is the degree and direction in which two variables are related or associated. Therefore, correlation coefficients are used to measure the strength of the relationship between two variables.

Let me tell you about them and their usefulness in #DataAnalytics 🧵

1/7

A correlation can be positive, meaning that the 2 variables move in the same direction, or negative, meaning that they move in opposite directions. When the correlation is neutral (close to 0) it means that the 2 variables are unrelated.

2/7

Two of the most important correlation coefficients are:
▪️ Pearson
▪️ Spearman

and what is the difference between them?

3/7

Pearson's correlation coefficient is a statistical measure of the strength of the linear relationship between two numerical variables.

It is a very useful coefficient , but it is also very sensitive to outliers.

4/7

Spearman correlation coefficient is a statistical measure of the strength of a monotonic relationship between two variables.

Spearman correlation is less sensitive than the Pearson correlation to outliers.

5/7

You must have used the border-radius property in CSS, but here is complete guide covering some of the hidden concept of it

🧵👇🏻

The border-radius CSS property rounds the corners of an element's outer border edge. You can set a border-radius in order to make round edges of sharp cornered box

Well we know that every element is a box in webpage. Consider that box clipped using 8 clips 👇🏻

We can adjust that clips in order to make distorted shapes

We can shift the position of eight clips and make some amazing random shapes.

The syntax is pretty easy, we can consider it as border-radius accepts 8 value for each clip, four value before slash (/) and four values after slash

Let's decode what each value means.

border-radius: 2px 4px 6px 8px / 10px 12px 20px 16px;

The first 2px simply means the distance of first clip from the top left corner and second value that is 4px means the distance of second clip from top right corner

Today on #KashmiriHinduExodus_31yrs ,i'd like to share my family's exodus story.We(Joint Family)were from rural Kashmir(Bidder,Kokernag) & my Parents had Govt Jobs in Srinagar & had rented a room at a Kashmiri Pandit’s(who had daughters)house at Sathoo Barbarshah,Wazir Lane.(1/n)

My Parents use to visit village every weekend to look after their 1 year old daughter.Just few days before 19th Jan 1990,my great grandfather passed away & my Parents had to leave Srinagar & stay in village for 10-15 days.When they came back to Srinagar,they were astonished (2/n)

to see dat their room owner has locked his house & left Kashmir for forever wid all of my Parents’ belongings still inside their rented room. My Parents enquired Owner’s ‘Peaceful’ neighbours about his whereabouts & as usual they told,”V don’t know why they've left Kashmir.”(3/n)

Somehow,my Parents arranged House’s Keys from Owner’s Relative & from that relative,my Parents finally came to know what actually happened on 19th Jan(e.g. slogans like,”We(Peacefuls) will make Kashmir Pakistan with Kashmiri Hindu Women,but without Kashmiri Hindu Men).Now as(4/n)

soon as my Parents entered inside their rented room & switched on d lights in evening,Local Peaceful’s network got alerted & started wondering,”How come this house’s owner has dared to come back to Kashmir after 10-15 days”.Just after 15-20 minutes,my Parents started hearing(5/n)

How Saif Ali Khan's great grandfather tried to merge Bhopal in Pakistan: [Thread]

Saif's great grandfather was the Nawab of Bhopal.

He was a staunch supporter of Pakistan & Jinnah. He governed Bhopal by Sharia.

He wrote a letter inviting King of Afghanistan to invade India.

August 2 1947: Saif Ali Khan's great grandfather, Nawab of Bhopal wrote a letter to Jinnah.

He clearly states

"I am a staunch supporter of Pakistan. I want to merge Bhopal in Pakistan. Hindus are untrustworthy.

But Bhopal is 80% Hindu and surrounded by India. I am cut off"

He further says

"Hindus are intimidating me to join India. I know Pakistan has no means of helping me. But I want to know how much Pakistan can help me.

I would never sign and join India, come what might . I will continue to serve Pakistan and Islam!"

In other words, Saif Ali Khan's family tried everything in their power to merge Bhopal in Pakistan. But as Bhopal was in the center of India, neither Jinnah nor Nawab had military means to resist India. Had Bhopal been a frontier kingdom, it would have been a part of Pakistan!

Source of the snippets: Quaid-i-Azam Mohammad Ali Jinnah Papers. From Jinnah's letter collection in Lucknow AIML office. Digitized and published in 1994-95 by Government of Pakistan and recommended for inclusion in Memory of World Register

Barlow Twins: a new super-simple self-supervised method to train joint-embedding architectures (aka Siamese nets) non

Basic idea: maximize the normalized correlation between a variable in the left branch and the same var in the right branch, while making the normalized cross-correlation between one var in the left branch and all other vars in the right branch as close to zero as possible.
2/N

In short: the loss tries to make the normalized cross-correlation between the embedding vectors coming out of the left branch and the right branch as close to the identity matrix as possible.
3/N

The 2 branches are always fed with differently-distorted version of the same image, and there is no need for dissimilar training pairs.

The objective makes the embedding vectors of the two branches as similar as possible, while maximizing their information content.
4/N

No contrastive samples, no huge batch size (optimal is 128), nor predictor, no moving-average weights, no vector quantization, nor cut gradients in one of the branches.
5/N