BC TECH

Saved by @Isaaciretai

Santiago
@svpino 5 years, 8 months ago 1225 views

Save to PDF Share See On Twitter

Here is a simple example of a machine learning model.

I put it together a long time ago, and it was very helpful! I sliced it apart a thousand times until things started to make sense.

It's TensorFlow and Keras.

If you are starting out, this may be a good puzzle to solve.

The goal of this model is to learn to multiply one-digit numbers.

https://t.co/w8LiMVENT0

It is a good example of coding, what is the model?
— Freddy Rojas Cama (@freddyrojascama) February 1, 2021

More from Santiago

Santiago
@svpino

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 😉.

Santiago
@svpino

10 machine learning questions you should be able to answer.

I started experimenting with a questionnaire before interviewing candidates. Applicants submit a video or written answers to apply for the position.

Thread: Let's look at 10 of these questions.

The questionnaire solves a couple of problems:

1. Weeds out candidates that aren't ready for the job.

2. It significantly cuts down the time we spend talking synchronously.

One problem: It requires extra time from the candidate before the phone interview.

Here are some numbers on a job posting before using the questionnaire:

• 249 applicants
• 13 good applications

That's only 5%.

This sucks.

Numbers after using the questionnaire:

• 44 applicants
• 21 good applications

That's 47%.

This is much better.

I look at it this way:

Sooner or later, I'll ask these questions, and the candidate will have to answer them.

I'd rather give them space and the opportunity to think about their answers without any pressure.

Those who can't or don't want to answer won't apply. That's fine.

Advertisement

Santiago
@svpino

My recommendation to start learning how to deal with structured data: Kaggle's Pandas Tutorial.

• Creating, Reading, Writing
• Indexing, Selecting, Assigning
• Summary Functions, Maps
• Grouping, Sorting
• Data Types, Missing Values
• Renaming, Combining

100% Python 🐍.

Santiago
@svpino

Do you know what's better than a machine learning model?

Two models.

More than one model working together to solve a problem is called "an emsemble." A simple way to build this is having each model vote for an answer.

But there's a problem with this approach: ↓

I'm gonna focus here on image classification.

Let's assume you built two different models:

• Model 1: A ResNet model.
• Model 2: A one-shot model (Siamese network.)

They both solve the same problem, so you want to combine their results to pick the right answer.

The problem is that you have two models, so voting is not trivial.

What happens in this case?

• Model 1's answer: Class A
• Model 2's answer: Class B

Which one do you select?

Notice that this problem is not limited to an even number of models.

You could have 3 models, each giving you a different answer.

How do you decide which answer to choose?

There are multiple ways to approach this problem. I'll mention a few different ideas on this thread.

Important: Some of these ideas might not be feasible depending on your context. They have worked for me before on different situations, but every problem is different.

Santiago
@svpino

20 machine learning questions that will make you think.

(Cool questions. Not the regular, introductory stuff that you find everywhere.)

🧵👇

1. Why is it important to introduce non-linearities in a neural network?

2. What are the differences between a multi-class classification problem and a multi-label classification problem?

3. Why does the use of Dropout work as a regularizer?

👇

4. Why you shouldn't use a softmax output activation function in a multi-label classification problem when using a one-hot-encoded target?

5. Does the use of Dropout in your model slow down or speed up the training process? Why?

👇

6. In a Linear or Logistic Regression problem, do all Gradient Descent algorithms lead to the same model, provided you let them run long enough?

7. Explain the difference between Batch Gradient Descent, Stochastic Gradient Descent, and Mini-batch Gradient Descent.

👇

8. What are the advantages of Convolution Neural Networks (CNN) over a fully connected network for image classification?

9. What are the advantages of Recurrent Neural Networks (RNN) over a fully connected network when working with text data?

👇

More from Tech

Michael Dearing
@mcgd

An irony of facebook's situation.

Sheryl got her MBA at Harvard. One of the most famous cases (Extra Strength Tylenol) in one of the most famous classes (Business History) she took: in 1982, someone put cyanide in Extra Strength Tylenol capsules and killed 7 people in Chicago.

What do you do when someone turns your product into a weapon? When they use the system you built to harm? James Burke, CEO of J&J, was shockingly open with the public, he pulled the product and made significant packaging changes to make product safer (but not tamper-proof).

He over-shared every step along the way re investigation, redesign, stood up as both CEO and human. The reintroduction of new Extra Strength Tylenol succeeded. Burke saved the brand.

But four years later it happened again. A killer put cyanide in the capsules, this time a woman in Yonkers died. Same CEO, Burke, pulled the product again, completely changed the form factor from capsule to caplet and relaunched *again*. It worked *again*. How'd they do that?

Burke (CEO) tapped J&Js goodwill bank account w/ the public. Two big withdrawals from that bank account in four years + 8 dead bodies! But his honesty, openness, humanity (choked up about the deaths more than once), humility kept the goodwill bank balance positive the whole time.

ashkan soltani
@ashk4n

BREAKING: @CommonsCMS @DamianCollins just released previously sealed #Six4Three @Facebook documents:

Some random interesting tidbits:

1) Zuck approves shutting down platform API access for Twitter's when Vine is released #competition

2) Facebook engineered ways to access user's call history w/o alerting users:

Team considered access to call history considered 'high PR risk' but 'growth team will charge ahead'. @Facebook created upgrade path to access data w/o subjecting users to Android permissions dialogue.

3) The above also confirms @kashhill and other's suspicion that call history was used to improve PYMK (People You May Know) suggestions and newsfeed rankings.

4) Docs also shed more light into @dseetharaman's story on @Facebook monitoring users' @Onavo VPN activity to determine what competitors to mimic or acquire in 2013.

https://t.co/PwiRIL3v9x

Advertisement

Jessie @ Chicago
@jessiedotjs

Why did I leave Google? A thread.

This thread is going to include quotes from Wired's "Three Years of Misery Inside Google, the Happiest Company in Tech" because I'm not a professional writer

When I joined Google, I fell in love with the company immediately. I was excited to come to work, not just for the technical challenges, but because I really felt I was making a difference and working for the good guys.

I spent 20% of my week working on an app that helped people find lost loved ones after natural disasters. I spent more free time speaking to students about technology and diversity. For an entire school year, I taught CS classes twice a week at a local low income high school.

The company was open. We had a weekly meeting called TGIF where anyone in the company could ask questions of, and raise concerns with, the execs. We got real answers.

El Sandifer, Rationali...
@ElSandifer

I’m quoted in this, which does about as good a job overviewing the topic as a New York Times profile can be expected to.

Major oversights in it as I see.

First, It does not acknowledge the fact that Yudkowsky is, at heart, a complete crank. This remains, to my mind, crucial to understanding the rationalist community’s influence on the world: they’re sci-fi writers being mistaken for scientists.

You can’t really understand the harmful effects of SSC until you realize that it’s part of a larger movement of bullshit artists serving as cult leaders to the techbros.

Second, and more importantly, it never really explains the harm that’s being done by Scott Siskind and his followers. It comes close, noting that “SJWs” were the only group not welcome in SSC, but it doesn’t capture the harm of that because it doesn’t unpack the term.

What is meant in practice when “SJWs” aren’t welcome in a space is that marginalized people speaking of the harm they experience are silenced.

That’s it. That’s the meaning.

Aviral Bhatnagar
@aviralbhat

Fintech firms will become digital layers around central banks, just like banks were physical layers

Central banks will create plug and play digital infrastructure (e.g. UPI, Aadhar) on which fintech products are built

Central banks will enable fintechs to replace banks

You May Also Like

James Felton
@JimMFelton

Butt plugs were originally sold as a miracle cure for headaches, acne, and insanity

Thread, article here and h/t: @qikipedia :
https://t.co/Ts84xnmWKJ

For a long time, medicine was pre-occupied with the question "can this be solved by putting something in the butt?"

Take, for example, tobacco smoke enemas, in which 18th Century physicians saw drowning victims and attempted to de-deadify them by blowing tobacco smoke up their rectums with a pipe.

Despite having no medical benefit whatsoever, the practice was popular, and enema kits lined the Thames like lifebelts, ready to be used as one final humiliation for somebody who was crap at swimming.

This practice ended sometime after 1800 when it became apparent that it didn't really work, given how the anus famously isn't really connected to the lungs, but doctors weren't quite finished with having a poke around in the old anus to see what happens just yet.

$Billy Bostickson \U0001f3f4\U0001f441&\U0001f441 \U0001f193$

Billy Bostickson 🏴👁&👁 ...
@BillyBostickson

Tip from the Monkey
Pangolins, September 2019 and PLA are the key to this mystery
Stay Tuned!

Meet Yang Ruifu, CCP's biological weapons expert https://t.co/JjB9TLEO95 via @Gnews202064
Interesting expose of China's top bioweapons expert who oversaw fake pangolin research
Paper 1: https://t.co/TrXESKLYmJ
Paper 2:https://t.co/9LSJTNCn3l
Pangolinhttps://t.co/2FUAzWyOcv pic.twitter.com/I2QMXgnkBJ
— Billy Bostickson \U0001f3f4\U0001f441&\U0001f441 \U0001f193 (@BillyBostickson) October 11, 2020

2. A jacobin capuchin dangling a flagellin pangolin on a javelin while playing a mandolin and strangling a mannequin on a paladin's palanquin, said Saladin
More to come tomorrow!

3. Yigang Tong
https://t.co/CYtqYorhzH
Archived: https://t.co/ncz5ruwE2W

4. YT Interview
Some bats & pangolins carry viruses related with SARS-CoV-2, found in SE Asia and in Yunnan, & the pangolins carrying SARS-CoV-2 related viruses were smuggled from SE Asia, so there is a possibility that SARS-CoV-2 were coming from

Advertisement

Mike DiCenzo
@mikedicenzo

"How the Avengers would respond to the Capitol riot" (a thread)

CAPTAIN AMERICA - tells Falcon, "I fought Nazis once before. They're extremely dangerous." They then see the guy scaling the Capitol wall fall into the shrubs. Falcon: "You sure these are the same Nazis?"

IRON MAN - lands by the guy carrying the podium. Says, "Mind if I borrow this?" then tosses it to Hulk, who swings it like a bat at the mob. Guy: "Hey, that's MY podium!" Iron Man: "Actually it's the Speaker's. And technically, it's a lectern." Then punches him through a wall.

BLACK WIDOW - enters hotel room where Jack Dorsey is tied up. Rips duct tape off his mouth. Jack: "I'll do anything you want!" BW: "Disable POTUS's Twitter. NOW!" Jack fumbles with his phone. "Done. Anything else?" BW: "Yeah. Trim your beard for God's sake." *tosses him clippers*

HULK - grabs a selfie stick from one of the guys and beats him over the head with it. Deeming it too small, he grabs another guy's Confederate flag and takes out a whole swath of rioters. He then tosses the flag to Cap, who uses it as a javelin to impale a fleeing Ted Cruz.

HAWKEYE - hears Iron Man in his earpiece say, "Looks like Orange Julius is tweeting from another account." Hawkeye: "On it." He locks in on Trump through the White House window and fires an arrow, knocking the phone out of Trump's hands. Hawkeye: "Damn that felt good."

Simon DeDeo
@SimonDeDeo

This is a pretty valiant attempt to defend the "Feminist Glaciology" article, which says conventional wisdom is wrong, and this is a solid piece of scholarship. I'll beg to differ, because I think Jeffery, here, is confusing scholarship with "saying things that seem right".

Imagine for a moment the most obscurantist, jargon-filled, po-mo article the politically correct academy might produce. Pure SJW nonsense. Got it? Chances are you're imagining something like the infamous "Feminist Glaciology" article from a few years back.https://t.co/NRaWNREBvR pic.twitter.com/qtSFBYY80S
— Jeffrey Sachs (@JeffreyASachs) October 13, 2018

The article is, at heart, deeply weird, even essentialist. Here, for example, is the claim that proposing climate engineering is a "man" thing. Also a "man" thing: attempting to get distance from a topic, approaching it in a disinterested fashion.

Also a "man" thing—physical courage. (I guess, not quite: physical courage "co-constitutes" masculinist glaciology along with nationalism and colonialism.)

There's criticism of a New York Times article that talks about glaciology adventures, which makes a similar point.

At the heart of this chunk is the claim that glaciology excludes women because of a narrative of scientific objectivity and physical adventure. This is a strong claim! It's not enough to say, hey, sure, sounds good. Is it true?

foone
@Foone

It is 2018 and this error message is a mistake from 1974.
This limitation, which is still found in the very latest Windows 10, dates back to BEFORE STAR WARS. This bug is as old as Watergate.

When this was developed, nothing had UPC codes yet because they'd just been invented.
Back when this mistake was made, There was only one Phone Company, because they hadn't been broken up yet. Ted Bundy was still on the loose. Babe Ruth's home run record was about to fall.

When this bug was developed, Wheel of Fortune hadn't yet aired. No one had seen Rocky Horror. Steven Spielberg was still a little-known directory of TV films and one box-office disappointment. SNL hadn't aired yet. The Edmund Fitzgerald was still hauling iron ore.

WHEN THIS STUPID MISFEATURE WAS INVENTED, THE GODFATHER PART II HAD JUST OPENED IN THEATERS.

So, why does this happen? So Unix (which was only 5 years old at this point) had the good idea of "everything is a file" which mean you could do things like write to sockets, pipes, the console, etc with the same commands and instructions.