Pascal Andre Dufour

Computer scientist, author, PhD. I write about practical #DeepLearning and #ComputerVision at

[AI Memes for Artificially Intelligent Teens @ai_memes]

Machine learning pipelines

[TensorFlow @TensorFlow]

🌟 The first release (0.1) of TFLite Support!

Check out this toolkit that makes deploying TFLite models on mobile devices much easier. It works cross-platform with Java, Python, C++, and Swift APIs.

Read more here → https://t.co/2iKAJcVr4E

[Raquel Adaia @adaiasand]

Check out my new post - Soldering for electrophysiologists!
https://t.co/9OP3x2oAQh
#neuroscience #invivoephys #methodsneuroscience #eegcommunity #rodentrecordings #electrophysiology

[Miles Cranmer @MilesCranmer]

Just learned about Python Fire, and wish I had heard about it years ago. Seems like an amazing library for productivity!

Fire turns any Python object—function, class, etc—into a command line interface:

Gone are the days of argparse.ArgumentParser and sys.argv...

[ySSN @y_SSN]

Are you neuro phd/postdoc in a Swiss university?
Are you motivated to organise and shape events for like-minded neuro-ERC folks?
If yes PM us, we are looking to get our team growing with volunteers! 💗🧠(please RT 🙏)

[Pascal Andre Dufour @dufourp83]

@madewithml Thanks for the interest! Hope you find it useful.

[Made With ML @MadeWithML]

🏆 Trending post of the day is Test-Time Data Augmentation by @dufourp83 - how to properly implement test-time image data augmentation in a production environment with limited computational resources.

https://t.co/9U5dwCXO9J

[Pascal Andre Dufour @dufourp83]

✨New tutorial is out ✨

Learn how to use test-time data augmentation (TTA) to improve predictions in a production environment with limited computational resources.

Experiment yourself with the @GoogleColab notebook with #keras and #tensorflow!

https://t.co/prkuXsMetX

[Elie Bursztein @elie]

Detecting deep-fake images is essential as they are increasingly weaponized. New research by @thorstenholz group show that using frequency analysis and deep-learning allows to detect deep-fakes generated with popular techniques reliably https://t.co/sAd8yLCVAU #DeepLearning #ai

[Angjoo Kanazawa @akanazawa]

New work in #ECCV2020 learning to reconstruct a morphable shape, texture, and viewpoint from an image collection without 3D ground truth *and* 2D keypoints, allowing us to explore new categories like shoes!
Congrats to @goelshbhm!
https://t.co/9fX482cgtp
https://t.co/bIL3RNJweC

[Pascal Andre Dufour @dufourp83]

New tutorial is out! ✨

Train a #DeepLearning classifier on CIFAR-10 and compare the performance with and without data augmentation.

Experiment for yourself with the @GoogleColab notebook!

https://t.co/X908II6jwS

[Pascal Andre Dufour @dufourp83]

Data augmentation is a powerful tool in #DeepLearning and important to master.

But what if the augmentation function you want to use does not exist yet?

Learn how to define your own in the @GoogleColab tutorial!

https://t.co/AxEjV46FJr

[AutoKeras @autokeras]

AutoKeras 1.0.4 is released! Important bug fixes and performance improvements has been made. Thank all the contributors! See detailed release notes below.
https://t.co/en2iq5LAfK

[Pascal Andre Dufour @dufourp83]

When you train a #DeepLearning model in #Keras, what parameters should you use for image data augmentation?

Find a visual exploration in the blog post and experiment for yourself on @GoogleColab!

https://t.co/FyHE2xuiTQ

[Pascal Andre Dufour @dufourp83]

✨Big News!✨

I started a #DeepLearning blog to share my knowledge!

Find concise, short tutorials on everything #AI and #MachineLearning.

Experimenting is key when learning, so all tutorials come with a @GoogleColab notebook!

https://t.co/LeR4XNcZ2c

[Pascal Andre Dufour @dufourp83]

@DeepMind lectures could be interesting for you if you're getting into #DeepLearning! Especially the more advanced lectures: generative and unsupervised methods, and #ResponsibleAI

https://t.co/nCV6CgEqt1

[Pascal Andre Dufour @dufourp83]

Bringing a #MachineLearning model to production is always hard. The slides by Luigiu Patruno neatly explain why.

#mlops https://t.co/WbVobI4w9c via @SlideShare

[Made With ML @MadeWithML]

👶 We are 4 months old today! 1.9K projects, 3K DAU, 5K MAU, 1.7K Slack (50K messages) & growing!

But what's MOST important to us is that the community has been building & we've become a place to showcase it! Check out a sample of the different kinds of portfolios 🔥

⬇️ Thread

[Robert Lange @RobertTLange]

There is a lot to wrap your head around in LSTMs🤯. One way of thinking that helped me a lot is the 'conveyor belt' metaphor of the cell state 🧑‍🏭 by @ch402. I put together a little animation 🖼️ Check out the amazing blog post by Chris Olah here✍️: https://t.co/BEXTawzsaE

[Richard Eudes, PhD @RichardEudes]

Top 8 Challenges for Machine Learning Practitioners https://t.co/Cvep9WiW7Y #analytics #datascience, #bigdata, #datascience, #datascience #ds, #machinelearning