Discover and read the best of Twitter Threads about #deeplearning

Most recents (24)

Why does every beginner data scientist fall for the "deep learning trap"?

True story 🧵

#rstats #datascience #deeplearning
When I was first learning data science this cost me at least 6-months. Seriously...

I was building a model for predicting which quotes would become orders.
I had just finished using a linear regression (didn't know about logistic yet) to make a predictive model.

Yeah I know - I was a noobie using regression instead of classification. So what?!
Read 15 tweets
What does a data science experiment look like and how do you design the first round of experiments?
#DataScience #MachineLearning #DeepLearning
Data Science experiments start with a question. How does the business keep customers subscribed or maintain their current level of spending?

Notice this is not the standard definition of churn and there’s a good reason for that.
#DataScience #MachineLearning #DeepLearning
Take existing data, train models, and compare them (multiple single models and ensembles) to the EXISTING customer retention process’s performance. This is your first experiment! Why haven’t we used a test dataset?
#DataScience #MachineLearning #DeepLearning
Read 12 tweets
What do model testing and accuracy metrics really tell you about your model's performance in production? Very little.

Test sets provide baseline accuracy metrics telling you how well you've modeled the dataset. That's not enough.
#DataScience #MachineLearning #DeepLearning
Remember what models learn, a function. That function represents a system like a customer's reaction to an ad or a part of the supply chain.

A trained model learns a function that's only as complete as the training data.
#DataScience #MachineLearning #DeepLearning
If your data contains all the features and only the features impacting the system/behaviors you’re trying to model
Represents the full problem space, then the function your model learns will perform well in production.
#DataScience #MachineLearning #DeepLearning
Read 7 tweets
5 years ago, I proposed using #DeepLearning to predict response to immunotherapy in TCR-Seq.
Last year, we published #DeepTCR. Today, I am proud to share its application in cancer immunology, now out in @ScienceAdvances #ScienceResearch
This project started with a hypothesis that if a specific antigen-specific response was characteristic of responses to immunotherapy, then this should be reflected in the TCR repertoire/sequences.

After all, the TCR repertoire is ground truth for the antigen-specific response.
First, we expand the #DeepTCR framework by allowing featurization of the TCR in context of the HLA background it is found within, overcoming the challenge in making direct comparisons of TCR repertoire in HLA mismatched individuals. Image
Read 21 tweets
Every Data scientist should know this!!
🧵⬇️ Image
Top 10 Industries for Data Science :
1. Finance
2. Retail
3. Technically
4. Healthcare
5. Telecom
6. Pharma
7. Automation
8. Cyber security
9. Energy
10. Utilities
Top 10 Roles for Data Science:
1. Data Scientist
2. Machine Learning Engineer
3. Analyst
4. ETL Engineer
5. Data Engineer
6. Analytics Manager
7. Tableau Developer
8. Researcher
9. BI Analyst
10. AI engineer
Read 10 tweets
Took a face made in #stablediffusion driven by a video of a #metahuman in #UnrealEngine5 and animated it using Thin-Plate Spline Motion Model & GFPGAN for face fix/upscale. Breakdown follows:1/9 #aiart #ai #aiArtist #MachineLearning #deeplearning #aiartcommunity #aivideo #aifilm
First, here's the original video in #UE5. The base model was actually from @daz3d #daz and I used Unreal's #meshtometahuman tool to make a #metahuman version. 2/9 #aiartprocess
Then I took a single still frame from that video and ran it through #img2img in a local instance of #stablediffusion WebUI. After generating a few options I ended up with this image. 3/9 #aiartprocess Image
Read 9 tweets
------------------Feature Engineering----------------------

The success of all Machine Learning algorithms depends on how you present the data. Every model gets input data and gives us an output. When your goal is to get the best possible output from input,

You need to present the best data to the model. This is a problem that Feature Engineering solves. Feature Engineering refers to the process of using the domain of Knowledge to extract features from raw data.

In other words, Feature Engineering selects the most useful features from our raw data and presents them to our model, whereby we improve the performance of our model.

(hopefully, you get the point 😀).

Read 10 tweets
Let's assume you have three Features(age, height, salary) in your example.
The first feature varies from 1 to 90. The second one varies from 120 to 210 and the Third one varies from 1000 Euro to 4500 Euro.
As you can see the value of your features are in a different range. In this case, if you want to use gradient descent to find optimum parameters for your model( for instance linear regression), that leads to a slow speed of your model to converge. In this case,
you can utilize Feature Scaling to bring the value of features in a range from 0 to 1 depending on the Scaling technique, that you use. So you improve the speed of your model convergence.

Read 5 tweets
As I started to learn Data Science I didn't know what skills should I learn and where. That was a ton of content and I didn't know which one should I take. I have read more than a hundred articles and talk with some of my data scientist friend and gathered experience
during my journey. I want to share a roadmap and skills that you need as a junior Data Scientist and resources to learn.

1- Start with a language programming and best of all Python. You can learn Python from 3 resources.
Taking one of these courses is enough.

A- 2022 Complete Python Bootcamp From Zero to Hero in Python by Jose Portilla in Udemy.

Jose Portilla is my favorite instructor. This Course has a GitHub repo where you can access Codes there.…

Read 19 tweets
Using X/Y Plot in a local #stablediffusion WebUI to create contact sheets exploring the latent space for my previous post. As a photo/video person I'm trying to bring the #aiartprocess closer to the workflows I use with clients.1/7 #aiphotography #aifashion #aiart #aiartcommunity
I imagine if you're coming to AI art from a different set of creative fields and client expectations, you'd approach the latent space from a different collection of frameworks and processes. Very curious how different a UI/UX catering to painters could look. 2/7 #aiartprocess
After an exhaustive process of prompting, I use different combos of X/Y Plot to make variations. The image I posted looked at sampler method+step count. Still trying to figure out what flow makes sense for me. I'm sure I'll throw it out as soon as new stuff gets released lol 3/7
Read 7 tweets
Love the X/Y Plot feature in AUTOMATIC1111's #stablediffusion WebUI. Once I have an image I like, I can do a quick search of the latent space around it across different settings. 1/3 #aiart #ai #aiArtist #MachineLearning #deeplearning #aiartcommunity
Link to AUTOMATIC1111's WebUI Repo here. Still have to check out its inpainting/outpainting implementations plus a few other features. 3/3 #stablediffusion #aiart #aiartists #ai #aiArtist #MachineLearning #deeplearning #aiartcommunity…
Read 3 tweets
سونالی فوگٹ کا کیس جہاں بہت آفسوس ناک ہے
وہیں اس کیس میں خواتین کے لیے بھی بہت کچھ سیکھنے کے لیے ہے
خصوصا خودمختار صاحب جائیداد خواتین
سونالی کئی ایکٹر کے ساتھ ذاتی فارم ہاوس کی مالک تھی
بلاشبہ حسن کی دولت سے بھی مالا مال تھی
سیاست میں بھی تھی
سب کچھ پرفیکٹ تھا
پھر اک ٹکلا سدھیر اسکی زندگی میں آتا ہے
خود کو کونٹیکٹر کے طور پر انٹرڈیوس کرواتا ہے
آسٹریلیا میں اک کمپنی کا مالک بھی
یہ سب باتیں بعد میں جھوٹ نکلنے پر سونالی کے علم میں لائی جاتیں ہیں
وہ سب باتوں کو سرسری لیتی ہے
یہ پہلا ریڈ فلیگ تھا جو اسنے اگنور کیا
جو کہ نہیں کرنا تھا
سونالی بااختیار مظبوط عورت ہونے کے باوجود اسکے زیر اثر آتی گئی
اور وہ اسکا آستحصال کرتا گیا
سونالی اگر پہلے ریڈ فلیگ کو سیریس لیتی
تو آج شائد ذندہ ہوتی
قدرت اشارے دیتی ہیں بس ہم سمجھ نہیں پاتے

Read 4 tweets
A wonderful book titled 'Deep Learning and Linguistic Representations (2021)' by Prof. Shalom Lappin.


#DeepLearning #naturallanguageprocessing #linguistics #representations…
Prof. Lappin's webpage:…
@threadreaderapp please unroll!
Read 3 tweets
A very good paper I came across this morning by the @DeepMind researchers. For the past five years Transformers have been one of the most dominant approaches to Deep Learning problems, especially in the #NLP domain.

However, despite many interesting papers on the topic, and lots of good open code, there has been a noticeable lack of *formal* definition of what transformed are, especially on the level of pseudocode.

This paper aims to rectify that. It provides pseudocode for almost all major Transformer architectures, including training algorithms.

Read 5 tweets
📢 Very excited to see our work #cytoself now published in @naturemethods . #cytoself is a deep learning method for fully self-supervised protein localization profiling and clustering. Led by @liilii_tweet with @kchev @LeonettiManuel at @czbiohub…

Everything started with a conversation with my friend & colleague @LeonettiManuel who worked on building #OpenCellCZB — a map of the human proteome:

We wondered, can we use #DeepLearning to map the landscape of protein sub-cellular localization?

The problem with #images , compared to #sequences, is that it is unclear how to compare them. For example, how to estimate localization similarity from pairs of images of fluorescently labeled cells? With sequences we have algorithms and tools. But for images?

Read 18 tweets
(1/4) New release for fastai! 🎊🎉

Yesterday v5 of fastai (aka Practical Deep Learning for Coders) was released. The course by @jeremyphoward was recorded at the University of Queensland. 🧵

Images credit: course website and book
#DeepLearning #machinelearning #opensource
(2/4) The course is free and focuses on applying deep learning and machine learning models to practical problems for computer vision, NLP, etc using #Python and #PyTorch
(3/4) It covers topics such as:
✅ Introduction to neural network
✅ Natural language (NLP) models
✅ Computer vision models
✅ Random forecast ❤️
✅ Convolutions networks
✅ Data ethics
Read 4 tweets
The issue with #NP problems is Uncertainty. The greater amount of information about the problem, the less Uncertainty & therefore, the less Computational Complexity. In case of no Uncertainty, NP=P. #NPplusInformationEqualsP @ulisescortes @sierra_carles @vdignum @wooldridgemike
In practice, that's what #DeepLearning does: it gathers data of the problem but runtime is P.
In Research too: we publish about problems until we find a P solution and then we continue optimising the solution in case we need a trade-off PSPACE/PTIME. #AI #MachineLearning
Could we create a Problem Information System to reduce the complexity of Research in general?Could it be automated to the greatest extent? That would be a General Problem Solver. We can be greedy & blindly publish in words everything we might solve, or we could start making a PIS
Read 5 tweets
اگر آپکو اپنی بیوی بدصورت لگتی ہے اسکے چہرے پہ ، ہونٹوں ہاتھ اور بازووں پہ بال ہیں تو ان بالوں کو تھریڈنگ ویکسنگ کے ذریعے صاف کیا جاسکتا ہے
اگر بیگم کا چہرہ چکنا چمکتا محسوس نہیں ہوتا
تو چہرے کو نکھارنے کے لئے ہزار دو ہزار کا فیشل اور بلیچ ہوتا ہے

اگر بیوی کے چہرے پر رونق نہیں ہے تو دودھ دہی اور پھل کھلائیں
یقین کریں آپکی بیوی بھی دیکھنے لائق ہوجائیگی
اگر اتنی گنجائش نہیں تو یہاں وہاں منہ مارنے سے گریز کریں
کیونکہ جن چکنی چمیلیوں کو خوبصورت سمجھ کے انکی مدح سرائی کرتے ہیں
وہ بھی اتنے خرچوں کے بعد ہی دیکھنے لائق بنتی ہیں
Read 3 tweets
(1/3) Introduction to PyTorch! 🚀🚀🚀

If you are looking for a resource to start with #PyTorch, I recommend you check the PyTorch Lightning tutorial - 𝘐𝘯𝘵𝘳𝘰𝘥𝘶𝘤𝘵𝘪𝘰𝘯 𝘵𝘰 𝘗𝘺𝘛𝘰𝘳𝘤𝘩 by Philip Lippe. 🧵👇

#DeepLearning #Python #OpenSource Image
(2/3) The tutorial is the first one out of a sequence of tutorials focusing on different topics of neural network and deep learning models and their applications 🌈.

License: CC BY-SA Image
Read 3 tweets
(1/4) New release for SuperGradients 👇🏼

The SuperGradients is a Python library that provides a framework for training deep learning models with #PyTorch for any 𝐜𝐨𝐦𝐩𝐮𝐭𝐞𝐫 𝐯𝐢𝐬𝐢𝐨𝐧 tasks by @deci_ai 🧵

#python #opensource #deeplearning #MachineLearning
(2/4) The package enables you to import pre-trained SOTA models, such as object detection, classification of images, and semantic segmentation for videos and images.

See quickstart:…
(3/4) Version 2.0.0/1 main updates:
➡️ Integration with @weights_biases and @TheRealDagsHub
➡️ KD trainer and resnet50 recipe (81.92 % accuracy)
➡️ ViT models (Vision Transformer) - Training recipes and pre-trained checkpoints (ViT, BEiT).

𝐋𝐢𝐜𝐞𝐧𝐬𝐞: Apache-2 🌈
Read 4 tweets
(1/4) Introduction to Deep Learning course 📚 📕

If you are looking for an introductory course for deep learning, I recommend checking the Introduction to Deep Learning course by Prof. @rasbt from the @UWMSTP 🧵 🐱 👇🏼

#DeepLearning #PyTorch #DataScience #python
(2/4) As its name implies, the course focuses on the foundation of deep learning using #Python and #PyTorch 🚀.
(3/4) The course is 15 weeks long and covers the following topics:
✅ Introduction to deep learning
✅ Mathematical and computational foundations
✅ Introduction to neural networks
✅ Deep learning for computer vision and language modeling
✅ Deep generative models
Read 4 tweets
ریکھا سے کسی نے پوچھا
آپ کو عمر کے اس حصے میں تنہائی کا احساس تو ہوتا ہوگا؟
ریکھا مسکرا دیں
کچھ پل کے لیے اپنی کلائی میں پہنا کنگن گھمایا اور پھر جواب دیا
جن لوگوں نے زندگی کو سیلیبریٹ نہیں کیا ہوتا تنہائی تو ان کے اندر جنم لیتی ہے
زندگی کا ہاتھ تھام کر چلتے چلے
زندگی کا ہاتھ تھام کر چلتے چلے جانے والوں کے اندر زندگی سے گِلے نہیں ہوتے
صرف خوشیاں ہوتی ہیں
میں زندگی سے خوش ہوں
زندگی مجھ سے خوش ہے , اک ساتھ آئے تھے اک ساتھ چلے جائیں گے

Read 3 tweets
Neural Network with Flax 🚀🚀🚀

#Flax is a Google #OpenSource Python library for neural network applications for JAX 🌈. While most folks are familiar with Google #TensorFlow and #Keras, 🧵 👇🏼

#DataScience #deeplearning #DL #python Image
While most folks are familiar with Google TensorFlow and Keras, Flax is less known, but it is mainly used by researchers and engineers at Google.

One of the core use cases of this library is for #NLP #Transformers and image recognition applications Image
Among the Flax applications, you can find:
✅ Neural network API (flax.linen): Dense, Conv, Norm, Attention, Pooling, Cell, Dropout
✅ Utilities and patterns: replicated training, serialization and checkpointing, metrics, prefetching on device
✅ Educational examples Image
Read 5 tweets
Learn Data Science in 180 days🤑📈 and start your data science career.

Bookmark this thread

A thread🧵👇
First Month 🗓️
Day 1 to 15 - Learn Python for Data Science
Day 16 to 30 - Learn Statistics for Data Science
Second Month 🗓️
Day 31 to 45 - Explore Python Packages( Numpy, Pandas, Matplotlib, Seaborn, Scikit-Learn)
Day 16 to 30 - Implement EDA on real-world datasets.
Read 8 tweets

Related hashtags

Did Thread Reader help you today?

Support us! We are indie developers!

This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3.00/month or $30.00/year) and get exclusive features!

Become Premium

Too expensive? Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal Become our Patreon

Thank you for your support!