BC ALL
Saved by @ankitsrihbti

This is Karma. Karma is not a machine learning classifier πŸ•β€πŸ¦Ί

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 πŸ”–

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 β€œprobable cause on four legs”.

Let's see why is that πŸ‘‡
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... πŸ‘‡
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 πŸ”

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 πŸ€¦β€β™‚οΈ
3. Leaking Training Data 🚰

Another study found that the dogs learned to recognize the emotions of their handlers during tests. They felt that their human wanted them to find drugs in the specific test scenario, so they did.

The trainer leaked the ground truth during testing.
4. Overfitting ➿

Similar to the one above, in many cases, the dog saw that their handler wanted to find drugs in a car during a traffic stop. So it would raise an alarm.

The dog was rewarded before the car was actually searched! It found an easy signal giving it a reward.
Summary 🏁

It is fascinating how many problems there are with the sniffer dogs that are well known to machine learning engineers (and of course mathematicians). Some of them are even common sense...

Avoid these problems not only when training your model, but also in life πŸ˜ƒ
If you liked this thread and want to read more about self-driving cars and machine learning follow me @haltakov!

More from Vladimir Haltakov

Vladimir Haltakov
Vladimir Haltakov
@haltakov
Let's talk about a common problem in ML - imbalanced data βš–οΈ

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 πŸ‘‡


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? πŸ‘‡

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

β–ͺ️ Choose the right evaluation metric
β–ͺ️ Undersampling your dataset
β–ͺ️ Oversampling your dataset
β–ͺ️ Adapting the loss

Let's dive in πŸ‘‡

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:
β–ͺ️ Precision
β–ͺ️ Recall
β–ͺ️ 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 πŸ‘‡
ALL
Vladimir Haltakov
Vladimir Haltakov
@haltakov
AI Job Interviews - another good example of bias in ML πŸ€¦β€β™‚οΈ

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 πŸ€·β€β™‚οΈ

Let's unpack πŸ‘‡


The Setup πŸ”¬

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... πŸ‘‡

The Positive Test βœ…

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

Then she tried something different... πŸ‘‡

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... πŸ‘‡

What happened? πŸ”

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.

πŸ‘‡
ALL
Advertisement
Vladimir Haltakov
Vladimir Haltakov
@haltakov
Machine Learning Paper Reviews πŸ”ŽπŸ“œ

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.

πŸ‘‡ I'm adding new reviews all the time! πŸ‘‡

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
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? πŸ€”

Let me shortly explain πŸ‘‡


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... πŸ‘‡

The ML model learned to detect the rulers... πŸ€·β€β™‚οΈ

"In our dataset, images with rulers were more likely to be malignant; thus the algorithm inadvertently β€œlearned” that rulers are
ALL
haltakov.eth \U0001f30d \U0001f1fa\U0001f1e6
haltakov.eth 🌍 πŸ‡ΊπŸ‡¦...
@haltakov
Machine Learning in the Real World 🧠 πŸ€–

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 πŸ‘‡

#RepostFriday


Collect Data πŸ’½

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.

πŸ‘‡

Traffic Lights 🚦

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

β–ͺ️ Lighting conditions
β–ͺ️ Weather conditions
β–ͺ️ Distances and viewpoints
β–ͺ️ Strange variants

And if we sample only 🚦 we won't detect πŸš₯ πŸ€·β€β™‚οΈ

πŸ‘‡


Data Cleaning 🧹

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

β–ͺ️ Overexposed or underexposed images
β–ͺ️ Images in irrelevant situations
β–ͺ️ Faulty images

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

πŸ‘‡


Preprocess Data βš™οΈ

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...)

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

πŸ‘‡
ALL

More from All

Nikita Poojary
Nikita Poojary
@niki_poojary
Catch me if you can @Mitesh_Engr

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

In collaboration with @AdityaTodmal


Edward de Bono says intelligence is like aπŸš— Automobile

β€œA 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:

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


How to Select Strikes to Sell?

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


Using Ratios to Sell:

β€’ If BNF is in a range then strangles
β€’ If breakout/ breakdown from a range then directional
β€’ Selling 70/30 or 80/20 ratios
β€’ 2 examples on how he used ratios
ALL , MITESH PATEL OPTIONS STRAT , MITESH PATEL EXPIRY GUIDE
Vibhu Vashisth
Vibhu Vashisth
@VIBHU_Tweet
RANI NAIKI DEVI AND THE BATTLE OF KASAHRADA

Mohammad Ghori, the young Ghurid Prince conquered Ghaznavids in Afghanistan in the year 1173 CE. He conducted some successful raids into the Indian territory, which even Mahmud Ghazni & the Persians couldn't do.


First&foremost,Ghori invaded states of Multan &Uch in India.After this he turned to southern states of Gujarat &Rajputana.Everybody knows this about Ghori,but what most pple r unaware of is the defeat he suffered at hands of Chalukyan Queen Naiki Devi in 1178 Battle of Kasahrada.


When he marched southwards, his first target was the prosperous fortified town of Anhilwara Patan.
It was the capital of Chalukyas(also known as Solankis of Gujarat) which was established by Vanraj of Chapotkata Dynasty in 8th century.


American historian Tertius Chandler states that it was the 10th largest city in the world at that time. The history of Chalukyas has been written by many historians, but the valour of the Great Queen Naiki Devi remains unsung.


Born to the Kadamba King Paramardin of Goa,Naiki Devi was well trained in sword fighting, horse riding, strategy & every other subject of Statesmanship. She was married to King Ajaypala of Gujarat who sat on throne in 1171CE. Unfortunately,the King died within 4 yrs of his rule.
ALL
Advertisement
Khatvaanga
Khatvaanga
@khatvaanga
Starting my 2nd thread on Sringeri Parampara. This will cover the Jagadgurus after Sri Vidyaranya to Sri Shivabhinava Nrisimha Bharati.

It is really very tough to match the brilliance and achievements of Vidyaranya. After whom came 15 - Chandrashekara Bharati [1386-1389].

The next Acharya was Nrisimha Bharati [1389-1408]. Harihara raya found a town for this Acharya - Narasimhapuram. Acharya helped the king by guiding him in political affairs. He also initiated the kind with mahamantras for spiritual development.

Subsequent Acharyas were
15-Puroshottama Bharati [1408 - 1448]
16-Shankara Bharati[1448 - 1455]
17-Chandrasekhara Bharati II [1455 - 1464]
18- Nrisimha Bharati II [1464 - 1479]
19-Puroshottama Bharati II [1479 - 1517]

20-Ramachandra Bharati [1517 - 1560]. Sadashiva Raya was emporer of Vijayanagara during this time.
21-Nrisimha Bharati III [1560 - 1573]
22-Nrisimha Bharati IV [1573 - 1576]
23-Nrisimha Bharati V [1576 - 1600]

24-Abhinava Nrisimha Bharati [1600-1623]. When he visited Malahanikareswara temple for first time he noticed there was no Vinayaka there so he took piece of turmeric and drew outline of Ganesha on a pillar and offered puja.
ALL
Angela Jamison
Angela Jamison
@AnJamison
@45RapeKatieJohn @buzz\_chronicles @buzz\\\_chronicles @buzz_chronicles save as category @pdfmakerapp grab @pdfmakerapp grab this @readwiseio save @readwiseio save thread @threader compile summarize @Threadreaderapp unroll please @ttttreads unroll @unrollthread.com @SaveToBookmarks
ALL
Buzz Chronicles
Buzz Chronicles
@buzz_chronicles
@SaveToBookmarks Ciao! There you go! Original tweet by @AchuthArora: "@powerbeastt @buzz_chronicles @rattibha @SaveToBookmarks save as Movies". See You soon! ✌

Ahoj! There you go! Original tweet by @SaveToBookmarks: "@AchuthArora @powerbeastt @buzz_chronicles @rattibha Cool!

I saved this tweet f...". Enjoy πŸ‘Œ
ALL

You May Also Like

Vikrant
Vikrant
@Trading0secrets
Trading view scanner process -

1 - open trading view in your browser and select stock scanner in left corner down side .

2 - touch the percentage% gain change ( and u can see higest gainer of today)


3. Then, start with 6% gainer to 20% gainer and look charts of everyone in daily Timeframe . (For fno selection u can choose 1% to 4% )

4. Then manually select the stocks which are going to give all time high BO or 52 high BO or already given.

5. U can also select those stocks which are going to give range breakout or already given range BO

6 . If in 15 min chartπŸ“Š any stock sustaing near BO zone or after BO then select it on your watchlist

7 . Now next day if any stock show momentum u can take trade in it with RM

This looks very easy & simple but,

U will amazed to see it's result if you follow proper risk management.

I did 4x my capital by trading in only momentum stocks.

I will keep sharing such learning thread 🧡 for you πŸ™πŸ’žπŸ™

Keep learning / keep sharing πŸ™
@AdityaTodmal
EQUITY , TRADINGTHREAD , TRADINGVIEW
Jaya_Upadhyaya
Jaya_Upadhyaya
@Jayalko1
JWALAMUKHI TRIPURASUNDARI, UTHANAHALLI, MYSURU (KAR)
#bharatmandir

Jwalamukhi Tripursundari is believed to be the sister of Devi Chamundeshwari, the presiding deity of Mysuru
It is also that she represents the flaming mouth of Jalandhara, the demon whom Shiva crushed to death.


When Devi Chamundeswari was fighting Mahishasura, she observed that from every drop of blood that spilled out of Mahishasura, another demon was born. So she created Tripurasundari to ensure that no blood dripped from Mahishasura.


Tripura Sundari means beauty of the three worlds.
The murti was installed during Wodeyar rule. An utsava murti made of pancha loha is kept in the sanctum sanctorum.
Another shrine here is dedicated to Siddeshwara / Shiva surrounded by a beautiful wooden Mantapa.
@GunduHuDuGa
ALL
Advertisement
\u0100yudhika
Δ€yudhika
@Satyamev1310
Ashta Siddhi

Ashta Siddhi is mentioned in Patanjali Yog sutra and can be obtained by very few through countless years of practice and devotion to all eight limbs of Ashtanga Yoga.

Siddhi can loosely be translated as accomplishment or a special power or unusual skill.


The eight Siddhis are:

1. Anima: Reducing one’s physical self to the size of an atom.

2. Mahima: Expanding one’s physical self to infinite level

3. Garima: Making one’s physical self so heavy as immovable by others

4. Laghima: Becoming almost weightless


5. Prapti: Being able travel anywhere in the Universe

6. Prakamya: Can know & understand what is going on in one’s mind. Can understand other person’s min.

7. Ishitva: Possessing title of Bhagwan

8. Vashitva: A seeker can subdue any person/thing

(Hanuman Jayanti is coming)


Hanumanji not only possessed all eight siddhis, but was also blessed by Sita mata as β€œAshta Siddhi Nau Nidhi Ke Daata”.

When Hanumanji first reaches to Lanka on his mission to find Sita, he makes use of anima so as to be discrete in enemy territory.


He uses it again when he approaches Sita mata for the first time – reducing his size to that of a schoolboy, to avoid scaring her. He uses mahima to outwit & overpowers
the demons.
ALL
Anu Satheesh
Anu Satheesh
@AnuSatheesh5
Sri Mridanga Saileswari Temple Muzhakkunnu, Kannur KERALA

The temple is established by Parasurama and it is one among 108 Durga temples.
Devi will answer all the prayers of her devotees who pray to her with utmost devotion lighting the ghee deepam.


It is the family diety of Kerala Varma Pazhassi Raja.
Devi is present in three forms Kali, Saraswati and Lakshmi.

Legend...πŸ‘‡
Once a Mridungam ( musical instrument) shaped rock fell from heaven here.
@SriramKannan77


Devi originated in the form of Mridangam and Parasurama captivated the essence of Devi into a Vigraham. Vigraham is made of Panchaloham.
Kathakali, an art form of Kerala originated here.


Temple Vigraham was stolen 3-4 times, but because of power of Devi all the attempts were failure. Police records says that Whenever the thieves try to steal, they lose their senses, and they will be in a confused state. So they will abandon the idol.
πŸ•‰
ALL
Vibhu Vashisth
Vibhu Vashisth
@VIBHU_Tweet
ANCIENT BHARAT WAS CALLED VISHWAGURU, AND IT REALLY DESERVED IT.

India had a well developed education system centuries before the westerners arrived and called us uncivilized. Education was always given a great importance in Indian civilisation since times immemorial.


Ancient India had the Gurukulas and Ashramas as the epicentres of the knowledge and enlightenment. Bigger Gurukulas served as the centres of higher education called universities. Besides these universities, temples also emerged as the major learning centres.

Studying the Holy Scriptures, character building, personality development, responsibilities towards self,family and society,discipline and preservation of the ancient culture and heritage were the key embodiments of education.

This kind of education system made ancient India, a centre of knowledge all over the world. Many foreign students came to India for education and India was called the 'Vishwaguru'.
Takshshila and Nalanda were the two prominent universities of ancient India.


But were these only Universities of ancient India?The answer is 'No'.Let's learn about these gems of our education system that flourished across ancient India & be proud.

NALANDA UNIVERSITY
This was the oldest university of ancient India.Situated in Nalanda distt. of Bihar...
ALL