Interesting paper bout reproducibility in AI/ML from Dr. Edward Raff is a Chief Scientist at Booz Allen Hamilton. He analyzed 255 papers, and successfully reproduce 162 from them.
A 62% success rate is higher than many meta-analyses from other sciences, and I suspect my 62% number is lower than reality
Interesting facts:
1. Having fewer equations per page makes a paper more reproducible.
2. Empirical papers may be more reproducible than theory-oriented papers.
3. Sharing code is not a panacea
4. Having detailed pseudo code is just as reproducible as having no pseudo code.
5. Creating simplified example problems do not appear to help with reproducibility.
6: Please, check your email (papers of people who answer on emails is more reproducible)
https://thegradient.pub/independently-reproducible-machine-learning/
A 62% success rate is higher than many meta-analyses from other sciences, and I suspect my 62% number is lower than reality
Interesting facts:
1. Having fewer equations per page makes a paper more reproducible.
2. Empirical papers may be more reproducible than theory-oriented papers.
3. Sharing code is not a panacea
4. Having detailed pseudo code is just as reproducible as having no pseudo code.
5. Creating simplified example problems do not appear to help with reproducibility.
6: Please, check your email (papers of people who answer on emails is more reproducible)
https://thegradient.pub/independently-reproducible-machine-learning/
The Gradient
Quantifying Independently Reproducible Machine Learning
Many warn that Artificial Intelligence has a serious reproducibility crisis, but is it so? Some conclusions from the author's experience trying to replicate 255 papers.
March 3, Kyiv
Data Science Meetup
Meet speakers:
- Nazar Shmatko, VP of engineering, RefaceAI
Topic: The success of generative models and how to reach it
- Philip Shurpik, Head of ML production, RefaceAI
Topic: ML & Video Pipelines - path to scalable production
Register for meetup: https://data-science.com.ua/en/events/data-science-meetup-2
Promocode: ML_World
Data Science Meetup
Meet speakers:
- Nazar Shmatko, VP of engineering, RefaceAI
Topic: The success of generative models and how to reach it
- Philip Shurpik, Head of ML production, RefaceAI
Topic: ML & Video Pipelines - path to scalable production
Register for meetup: https://data-science.com.ua/en/events/data-science-meetup-2
Promocode: ML_World
Data Science UA
Data Science Meetup #2 - Data Science UA
Hey guys, on this Saturday at Data Science UA Conference
I'm making a workshop about Object Detection with Single Shot Networks in PyTorch
You can register here https://bit.ly/2Np7VGy
10% off with code - ML_World
PS: Don't forget your masks for additional safe
I'm making a workshop about Object Detection with Single Shot Networks in PyTorch
You can register here https://bit.ly/2Np7VGy
10% off with code - ML_World
PS: Don't forget your masks for additional safe
Interesting paper about motion duplication for videos (aka making animation from image)
https://aliaksandrsiarohin.github.io/first-order-model-website/
https://aliaksandrsiarohin.github.io/first-order-model-website/
Дмитрий Меньшиков запустил канал по фин. грамотности, где расказывает как эффективно хранить и инвестировать деньги, как легально открыть счет в иностранном банке и прочее. Вообщем рекомендую!
https://tttttt.me/FinVam
https://tttttt.me/FinVam
Quarantine is the best time to learn something new. 20% discount on the practical online course - Object Detection with PyTorch.
https://learnml.today/c/covid19
https://learnml.today/c/covid19
learnml.today
Learn ML Today - Object Detection with PyTorch Course
This course is designed by Machine Learning Engineer with the aim to create experts in Object Detection. You will build complex models by 'learn by doing' style through the applied theme of Advanced Computer Vision Techniques.
Hey fellas!
For those who're bored at home, for those who want to see friends and talk to them, the chance is right here.
Let's meet at ML meetup, which will take place in virtual reality. All you need is a computer, VR-headsets aren't required. You'll log in, set up the avatar, select the seat, and will appear in the conference room, surrounded by all the folks from regular ML meetups. You'll be able to start a tet-a-tet chat, take photos/videos, hang out with friends during the afterparty and do some stuff from real life.
https://www.facebook.com/events/1871332413001282/
WHO'S GONNA SPEAK?
👨💻 Stepan Maksymchuk - CD4ML using Kubeflow and GitHub Actions
Continuous Delivery for Machine Learning (CD4ML) is a software engineering approach in which a cross-functional team produces machine learning applications based on code, data, and models in small and safe increments that can be reproduced and reliably released at any time, in short adaptation cycles.
Meet Stepan Maksimchuk from Ventus Cloud AG and learn how GitHub Actions and Kubeflow can be used together to implement the full end-to-end process for a sample ML application.
👨💻 Paul Galushko - What an ML engineer should know to deploy successful services to production
How can an ML model remain useful without becoming a mathematical puzzle? We’ll talk about ravines that are not obvious from the jupyter notebook.
For ML to become a real service, we’ll cover the following:
- Production hardware
- Cost of the service
- Deployment of ML and services, their delivery to users
- Mass re-learning of models, if their number exceeds 3
- Common issues for high-load services
- Design principles of industrial development, that help you live in harmony with yourself, your code, your colleges, and not to be afraid to go out even after quarantine.
https://www.facebook.com/events/1871332413001282/
For those who're bored at home, for those who want to see friends and talk to them, the chance is right here.
Let's meet at ML meetup, which will take place in virtual reality. All you need is a computer, VR-headsets aren't required. You'll log in, set up the avatar, select the seat, and will appear in the conference room, surrounded by all the folks from regular ML meetups. You'll be able to start a tet-a-tet chat, take photos/videos, hang out with friends during the afterparty and do some stuff from real life.
https://www.facebook.com/events/1871332413001282/
WHO'S GONNA SPEAK?
👨💻 Stepan Maksymchuk - CD4ML using Kubeflow and GitHub Actions
Continuous Delivery for Machine Learning (CD4ML) is a software engineering approach in which a cross-functional team produces machine learning applications based on code, data, and models in small and safe increments that can be reproduced and reliably released at any time, in short adaptation cycles.
Meet Stepan Maksimchuk from Ventus Cloud AG and learn how GitHub Actions and Kubeflow can be used together to implement the full end-to-end process for a sample ML application.
👨💻 Paul Galushko - What an ML engineer should know to deploy successful services to production
How can an ML model remain useful without becoming a mathematical puzzle? We’ll talk about ravines that are not obvious from the jupyter notebook.
For ML to become a real service, we’ll cover the following:
- Production hardware
- Cost of the service
- Deployment of ML and services, their delivery to users
- Mass re-learning of models, if their number exceeds 3
- Common issues for high-load services
- Design principles of industrial development, that help you live in harmony with yourself, your code, your colleges, and not to be afraid to go out even after quarantine.
https://www.facebook.com/events/1871332413001282/
10.1038@s41593-020-0608-8.pdf
2.7 MB
Machine translation of cortical activity to text with an encoder–decoder framework