BC ALL
Saved by @nandanarunesh

AI Job Interviews - another good example of bias in ML 🤦‍♂️

Two journalists tested some AI tools for assessing job candidates. Even when the candidate read a Wiki article in German instead of answering questions in English, the AI systems gave them good scores 🤷‍♂️

Let's unpack 👇

 The Setup 🔬

The journalists created a fake job posting on two AI interview platforms. They specified the traits of the ideal candidate and provided the questions that need to be answered during the interview.

Then they started experimenting... 👇
The Positive Test ✅

One of them did a fake interview giving all the right answers and predictably got very high scores - 8.5 out of 9 👍

Then she tried something different... 👇
The Negative Test ❌

In a second interview, instead of answering the questions in English, she just read the article on psychometrics from the German Wikipedia 😁

One system gave her a score of 6 out of 9, while the other determined she is a 73% match for the job.

Oops... 👇
What happened? 🔍

Interestingly, one of the systems generated a transcript which was obviously meaningless.

This means that the machine learning model behind the tool likely captured nuances of the intonation of the speaker instead of the meaning of the actual words.

👇
Bias ⚖️

These tools are biased - they incorrectly put more weight on the intonation instead of the actual meaning of what is being said.

This is shown when one of the components is artificially removed - in this case, the meaning (German instead of English).

More examples 👇
Bias again ⚖️

Another AI interview tool was reported to give candidates different ratings when they used different backgrounds or accessories during the interviews.

Why does this happen? 👇
The Dataset 💽

One likely reason is the dataset used to train the machine learning models. It may not contain enough data for the model to correctly learn to discriminate important features. It overfits on some non-essential features, like the background.

👇
Black-Box Models ⬛

Another common problem is the use of black-box models where it is difficult to interpret why the algorithm makes a certain decision (for example in the case of neural networks).

More work is needed to inspect what the model is paying attention to.

👇
Lack of Testing 🧪

Furthermore, it seems that the systems weren't tested in scenarios like this one and weren't prepared to meaningfully handle these cases. The question is then, what other cases will not be handled correctly?

👇
Bias in Human Interviews 👤

Interestingly, human interviews are also subject to many biases and this is a known problem. The hope is that AI systems will help reduce those biases, but one needs to be careful to not just replace them with different ones.
More Details 🗒️

For more details, you can read the full article by @HilkeSchellmann and @sheridanlwall on MIT Technology Review here:

https://t.co/uHAlrA2vKw
I regularly post similar threads on machine learning, computer vision, and self-driving.

Follow me @haltakov for more!
Yes, this is the exact goal of most of these AI tools - to provide another data point for interviewers to help them make a better decision.

https://t.co/Vy73ydcp61
Here is a very detailed article about the influence of different backgrounds and accessories on the scores

https://t.co/7hBveNTpHb https://t.co/gZkJ6zol9W

More from Vladimir Haltakov

Vladimir Haltakov
Vladimir Haltakov
@haltakov
Let's talk about a common problem in ML - imbalanced data ⚖️

Imagine we want to detect all pixels belonging to a traffic light from a self-driving car's camera. We train a model with 99.88% performance. Pretty cool, right?

Actually, this model is useless ❌

Let me explain 👇


The problem is the data is severely imbalanced - the ratio between traffic light pixels and background pixels is 800:1.

If we don't take any measures, our model will learn to classify each pixel as background giving us 99.88% accuracy. But it's useless!

What can we do? 👇

Let me tell you about 3 ways of dealing with imbalanced data:

▪️ Choose the right evaluation metric
▪️ Undersampling your dataset
▪️ Oversampling your dataset
▪️ Adapting the loss

Let's dive in 👇

1️⃣ Evaluation metrics

Looking at the overall accuracy is a very bad idea when dealing with imbalanced data. There are other measures that are much better suited:
▪️ Precision
▪️ Recall
▪️ F1 score

I wrote a whole thread on


2️⃣ Undersampling

The idea is to throw away samples of the overrepresented classes.

One way to do this is to randomly throw away samples. However, ideally, we want to make sure we are only throwing away samples that look similar.

Here is a strategy to achieve that 👇
ALL
Vladimir Haltakov
Vladimir Haltakov
@haltakov
Machine Learning Paper Reviews 🔎📜

Check out this thread for short reviews of some interesting Machine Learning and Computer Vision papers. I explain the basic ideas and main takeaways of each paper in a Twitter thread.

👇 I'm adding new reviews all the time! 👇

AlexNet - the paper that started the deep learning revolution in Computer Vision!


DenseNet - reducing the size and complexity of CNNs by adding dense connections between layers.


Playing for data - generating synthetic GT from a video game (GTA V) and using it to improving semantic segmentation models.


Transformers for image recognition - a new paper with the potential to replace convolutions with a transformer.
ALL
Advertisement
haltakov.eth \U0001f30d \U0001f1fa\U0001f1e6
haltakov.eth 🌍 🇺🇦...
@haltakov
Machine Learning in the Real World 🧠 🤖

ML for real-world applications is much more than designing fancy networks and fine-tuning parameters.

In fact, you will spend most of your time curating a good dataset.

Let's go through the process together 👇

#RepostFriday


Collect Data 💽

We need to represent the real world as accurately as possible. If some situations are underrepresented we are introducing Sampling Bias.

Sampling Bias is nasty because we'll have high test accuracy, but our model will perform badly when deployed.

👇

Traffic Lights 🚦

Let's build a model to recognize traffic lights for a self-driving car. We need to collect data for different:

▪️ Lighting conditions
▪️ Weather conditions
▪️ Distances and viewpoints
▪️ Strange variants

And if we sample only 🚦 we won't detect 🚥 🤷‍♂️

👇


Data Cleaning 🧹

Now we need to clean all corrupted and irrelevant samples. We need to remove:

▪️ Overexposed or underexposed images
▪️ Images in irrelevant situations
▪️ Faulty images

Leaving them in the dataset will hurt our model's performance!

👇


Preprocess Data ⚙️

Most ML models like their data nicely normalized and properly scaled. Bad normalization can also lead to worse performance (I have a nice story for another time...)

▪️ Crop and resize all images
▪️ Normalize all values (usually 0 mean and 1 std. dev.)

👇
ALL
Vladimir Haltakov
Vladimir Haltakov
@haltakov
Dataset bias strikes again! 😬

This is a great example of dataset bias! This ML method classifies skin lesions as being malignant or not.

Can you see the problem from the examples? 🤔

Let me shortly explain 👇


The dataset used to train the model contains images of both malignant and benign lesions. The problem is that the malignant images often contain rulers printed on the skin. They are used by doctors to assess the lesion over time.

Guess what the model learned... 👇

The ML model learned to detect the rulers... 🤷‍♂️

"In our dataset, images with rulers were more likely to be malignant; thus the algorithm inadvertently “learned” that rulers are
ALL
Vladimir Haltakov
Vladimir Haltakov
@haltakov
This is Karma. Karma is not a machine learning classifier 🐕‍🦺

Karma is a real dog trained to detect drugs. However, he would fail the simplest tests we apply in ML...

Let me take you through this story from the eyes of an ML engineer.

https://t.co/WAXRUlTvSI

Thread 🧵

Story TLDR 🔖

The story is about police dogs trained to sniff drugs. The problem is that the dogs often signal drugs even if there are none. Then innocent people land in jail for days.

The cops even joke about the “probable cause on four legs”.

Let's see why is that 👇

1. Sampling Bias 🤏

Drugs were found in 64% of the cars Karma identified, which was praised by the police as very good. In the end, most people don't carry drugs in their cars, so 64% seems solid.

There was a sampling problem though... 👇

The cars were not sampled at random! The police only did the sniff test if there was a serious suspicion that something is wrong.

The chance there are drugs in the car is much higher in this case!

2. Evaluation Metrics 🔍

The police referred to a 2014 study from Poland measuring the efficacy of sniffer dogs. The problem was that every test actually contained drugs!

This means there was no chance to measure false positives from the dogs! Only recall, not precision 🤦‍♂️
ALL

More from All

Aditya Todmal
Aditya Todmal
@AdityaTodmal
Master Thread of all my threads!

Hello!! 👋

• I have curated some of the best tweets from the best traders we know of.

• Making one master thread and will keep posting all my threads under this.

• Go through this for super learning/value totally free of cost! 😃

1. 7 FREE OPTION TRADING COURSES FOR


2. THE ABSOLUTE BEST 15 SCANNERS EXPERTS ARE USING

Got these scanners from the following accounts:

1. @Pathik_Trader
2. @sanjufunda
3. @sanstocktrader
4. @SouravSenguptaI
5. @Rishikesh_ADX


3. 12 TRADING SETUPS which experts are using.

These setups I found from the following 4 accounts:

1. @Pathik_Trader
2. @sourabhsiso19
3. @ITRADE191
4.


4. Curated tweets on HOW TO SELL STRADDLES.

Everything covered in this thread.
1. Management
2. How to initiate
3. When to exit straddles
4. Examples
5. Videos on
OPTION SELLING , OPTIONSTRATEGIES , CATEGORY
Vibhu Vashisth
Vibhu Vashisth
@VIBHU_Tweet
LACHIT BORPHUKAN : AN UNSUNG WARRIOR OF THE AHOM KINGDOM & THE GREAT BATTLE OF SARAIGHAT

Originally named as Lachit Deca, Lachit Borphukan, the fierce and invincible Ahom Commander, was born during the early 17th century at Betioni in the Golaghat district of modern Assam.


His father, Momai Tamuli Borbarua was the ‘Governor’ of the kingdom and also ‘Commander-in-Chief’ of Ahom army under King Pratap Singha during his reign starting from 1603 to 1639. Lachit received military training from an early age and joined the Ahom King Jayadhvaj Singha...


...(1648-1663) as a scarf-bearer.  The post ‘scarf-bearer’ or ‘Soladhara Barua’ is considered as a part of the incumbent king’s personal staff. Owing to his military training Lechit was appointed "Ghora Barua", Supdt. of the Royal horses. When Chakradhwaj Singha became king..


..of Ahom dynasty in 1663,he appointed him Suptd. of The Royal Guards &finally he was appointed as Borphukan by the king.Then onwards,he was known as Lachit Borphukan.Borphukan used to be top councillor in Ahom Kingdom. This position was embedded with executive & judicial powers.


Assam was facing incessant Islamic invasion since Jan1662, when Mughal General Mir Jumla II attacked Ahom capital Gargaon. He never defeated Ahom king Jayadhwaja Singha as king retreated to hill &continued Guerrilla warfare.Prolonged fight &prospect of stalemate disturbed Jumla.
ALL
Advertisement
Nikita Poojary
Nikita Poojary
@niki_poojary
Catch me if you can @Mitesh_Engr

Time for a🧵
Mitesh Sir's EXPIRY Option Selling 101:
• What to look for?
• Strike Selection & Ratios
• SL mgmt
• Avoiding freezes
• Monthy Expiry
• Event days
• How he would have traded last expiry?

In collaboration with @AdityaTodmal


Edward de Bono says intelligence is like a🚗 Automobile

“A powerful car may be driven badly. A less powerful car may be driven well. The skill of the car driver determines how the power of the car is used.”


What @Mitesh_Engr Sir looks at for Expiry Trading:

• Technicals of BNF
• Individual Heavyweight stocks of BNF
• Support/Resistance of BNF on weekly TF
• Also has Superb SL management
• When it comes to adjustments, he is always on toes


How to Select Strikes to Sell?

• You can't sell based on premium
• Follow only trend
• Price action is best to sell on expiry day
• Never wait for the options to go to zero


Using Ratios to Sell:

• If BNF is in a range then strangles
• If breakout/ breakdown from a range then directional
• Selling 70/30 or 80/20 ratios
• 2 examples on how he used ratios
ALL , MITESH PATEL OPTIONS STRAT , MITESH PATEL EXPIRY GUIDE
Andrew Lokenauth
Andrew Lokenauth
@FluentInFinance
I've used Microsoft Excel for 20 years and these 10 shortcuts will make you an Excel Guru in 5 minutes & maximize your productivity 100x:

🧵 Don't use Excel without knowing these 10 shortcuts:

(1) Repeat the Last Action: Ctrl + Y

You have just applied a formatting style to a cell, use Ctrl + Y to quickly apply the same formatting to another cell.


(2) Display Formulas in cells: Ctrl + ~

If you have a cell that contains a formula, you can use Ctrl + ~ to switch between the formula view and the value view of the cell. This allows you to see the underlying formula used to calculate the cell value.


(3) AutoSum: Alt + =

The AutoSum shortcut is useful for quickly calculating the sum of a range of cells without having to manually type in the formula


(4) Insert current date: Ctrl + ; (date)

When tracking the progress of a project, add the date each time a task is completed. Use Ctrl + ; to quickly insert the current date.
ALL
Save To Bookmarks \uea00
Save To Bookmarks ...
@SaveToBookmarks
@buzz_chronicles @AchuthArora @aldothetrainer @Artofphysique_ @rattibha Gotcha, boss!

Your tweet has been saved for later.

In order to manage your bookmarks you have to join up at savetobookmarks .com. Thanks!
ALL

You May Also Like

Sam Dylan Finch \U0001f937\U0001f3fb\u200d\u2642\ufe0f
Sam Dylan Finch 🤷🏻‍♂️...
@samdylanfinch
I had the most bizarre experience outside of a bar yesterday, which my partner very aptly described as "toxic mask-unlinity." (A Thread.)

I was standing outside, waiting for a lyft to get home, when two cis men came out of the bar to smoke cigarettes. They're discussing the masks that folks are wearing in California at the moment because of the very dangerous air pollution from the fires happening.

I happen to be wearing one of those masks, because I think lungs are important and I like breathing. I dunno. One of the men notices me standing there, and says, "Like this f*cker right here. What is this shit?"

He goes on a homophobic rant about me, the pansy with a mask, and what a Very Manly Man he is because he's not afraid to smoke a pack of cigarettes while California is on fire. He doesn't need a mask, because something-something-heterosexual-something.

And part of me is just... impressed by the fact that this dude thinks he's Owning The Gay Libs by not wearing a mask to protect his lungs from smoke and particles in the air that are literally DANGEROUS... while chain-smoking. Outside of a bar.
SOCIETY
Vibhu Vashisth
Vibhu Vashisth
@VIBHU_Tweet
हिमालय पर्वत की एक बड़ी पवित्र गुफा थी।उस गुफा के निकट ही गंगा जी बहती थी।एक बार देवर्षि नारद विचरण करते हुए वहां आ पहुंचे।वह परम पवित्र गुफा नारद जी को अत्यंत सुहावनी लगी।वहां का मनोरम प्राकृतिक दृश्य,पर्वत,नदी और वन देख उनके हृदय में श्रीहरि विष्णु की भक्ति अत्यंत बलवती हो उठी।


और देवर्षि नारद वहीं बैठकर तपस्या में लीन हो गए।इन्द्र नारद की तपस्या से घबरा गए।उन्हें हमेशा की तरह अपना सिंहासन व स्वर्ग खोने का डर सताने लगा।इसलिए इन्द्र ने नारद की तपस्या भंग करने के लिए कामदेव को उनके पास भेज दिया।वहां पहुंच कामदेव ने अपनी माया से वसंतऋतु को उत्पन्न कर दिया।


पेड़ और पौधों पर रंग बिरंगे फूल खिल गए और कोयलें कूकने लगी,पक्षी चहकने लगे।शीतल,मंद,सुगंधित और सुहावनी हवा चलने लगी।रंभा आदि अप्सराएं नाचने लगीं ।किन्तु कामदेव की किसी भी माया का नारद पे कोई प्रभाव नहीं पड़ा।तब कामदेव को डर सताने लगा कि कहीं नारद क्रोध में आकर मुझे श्राप न देदें।

जैसे ही नारद ने अपनी आंखें खोली, उसी क्षण कामदेव ने उनसे क्षमा मांगी।नारद मुनि को तनिक भी क्रोध नहीं आया और उन्होने शीघ्र ही कामदेव को क्षमा कर दिया।कामदेव प्रसन्न होकर वहां से चले गए।कामदेव के चले जाने पर देवर्षि के मन में अहंकार आ गया कि मैने कामदेव को हरा दिया।

नारद फिर कैलाश जा पहुंचे और शिवजी को अपनी विजयगाथा सुनाई।शिव समझ गए कि नारद अहंकारी हो गए हैं और अगर ये बात विष्णु जी जान गए तो नारद के लिए अच्छा नहीं होगा।ये सोचकर शिवजी ने नारद को भगवन विष्णु को ये बात बताने के लीए मना किया। परंतु नारद जी को ये बात उचित नहीं लगी।
ALL
Advertisement
TXgrlWatching\U0001f4bb\u2694\U0001f6e1
TXgrlWatching💻⚔🛡...
@VeritasTXgem
Thread A History lesson for the MOAB:
1. The United States of America is not and has never been a Democracy, we're a Federal Republic tied together by the capital Laws of the Constitution.

Everyone in the country is about to get a harsh lesson on what this really means...

2. In 1871, February 21: Congress Passes an Act to Provide a Government for the District of Columbia, also known as the Act of 1871.

With no constitutional authority to do so, Congress creates a separate form of
government for the District of Columbia.

3. A ten mile square parcel of land (see,
Acts of the Forty-first Congress,” Section 34, Session III, chapters 61 and

4. The act — passed when the country was weakened and financially depleted in
the aftermath of the Civil War — was a strategic move by foreign interests
(international bankers) who were intent upon gaining a stranglehold on the coffers and neck of America.

5. Congress cut a deal w/ the international bankers(Rothschilds of London) to incur a DEBT to said bankers. Because the bankers were not about to lend money to a floundering nation w/out serious stipulations, they devised a way to get their foot in the door of the United States.
HISTORYLESSON
Coffee Sloth \u2615
Coffee Sloth ☕
@CaseyExplosion
*Heads-up that this thread will feature some extreme examples of homophobia and transphobia.*

Hey folks, have you ever wondered why trans people face constant accusations of fetishism, sexual predation and child abuse/grooming? Well, today let's talk about stochastic terrorism.

So, first we have to talk about what stochastic terrorism actually is. Remember when there was a conspiracy that there was a pedophile ring operating out of a pizza parlour basement, and a crazed gunman showed up? It's that sort of incitement that stochastic terrorism describes.

It's the demonisation or incitement against a group of people or individuals with the intent that other, unaffiliated parties will act upon it; it's the releasing doctored footage and a shooter show up at Planned Parenthood as a direct result of what he's been led to believe.

So with that in mind, let's talk about Russia's extreme anti-LGBT laws that banned the "promotion" of LGBT lifestyles to children. A move that specifically marked LGBT people as a threat to children, and resulted in neo-nazis meting out vigilante "justice" against gay men.


Groups such as this would lure unsuspecting gay men via dating sites, and brutalise and humiliate them on camera. They did so under the guise of combating child abusers, one group calling themselves "Occupy
SOCIETY
foone
foone
@Foone
So there's a class of puzzles called "river crossing puzzles" which involve having to cross a river on a boat/bridge containing some set of individuals/items with restrictions on which ones can be on the boat/left on each shore, right?

and the most common subclass of these problems is the one where you have 3 items/animals to cross with.
This is alternatively called the "fox, goose, and bag of beans" or "wolf, goat, and cabbage" puzzle.

the idea is that your boat can only carry one item at a time, but if you leave the wolf/fox unattended with the goat/goose, the wolf/fox will eat the goat/goose.
and similarly, the goose/goat will eat the beans/cabbage if left alone

so you have to set up a set of actions that never leaves either side of the river in one of the lose-states.

the solution to this variant being:
1. take the goat/goose over
2. return empty
3. take the wolf/fox over
4. return with the goat/goose
5. take the cabbage/beans over
6. return empty
7. take the goat/goose over
GAMES