Released v0.2 of the Rust Language argminmax project! ✨ Some new features: * runtime CPU feature set selection 🤖 * float16 support (with #SIMD!) by using some cheeky bitwise operations => 300x faster than #numpy 🚀 github.com/jvdd/argminmax #rust #datascience

Paper published! We show that simple, #interpretable #machinelearning approaches compete with complex deep learning models for #sleepscoring 😴. Our method is easier to train, deploy, and interpret, making it a practical choice for automatic sleep scoring in clinical settings 🥂

🤝

🚨 New paper alert! 🚨 "Acoustic speech features in social comparison: how stress impacts the way you sound" in Scientific Reports together with Jonas Van Der Donckt nature.com/articles/s4159… Here's a 🧵on what we found! #OpenScience #AcademicChatter #phdchat 1/12

Humbled to see our work being shared by a renowned expert in the industry! 🥂 Our paper illustrates that a good feature vector is all you need ⬇️ #machinelearning #deeplearning #timeseries #sleep #opensource

Go check out our work that shows that you should not sleep on classical ML methods 👇! Classic ML is at least competitive to deep learning methods for sleep scoring.

I'm thrilled to announce that NaN handling has been added to the #rustlang argminmax project! 🚀 Even more exciting, this feature was the result of a collaborative effort within the #opensource community! 🤝 github.com/jvdd/argminmax… #FOSS #rust #simd #sse #avx #avx512 #neon

Exciting news! My first submission landed me in the top 3 of the #Parkinson's disease Kaggle competition 🏆 🚀 It's always fun to work on #timeseries #machinelearning ! #kaggle #datascience

🐐

Exciting news! Just made my first (major) contribution to polars data 🥳 I implemented the argminmax package to enhance the efficiency of the arg_min and arg_max operations! Chunked arrays saw a performance boost of up to 25x 🚀 Check out the PR here 👉 github.com/pola-rs/polars…

tsdownsample v0.1.2 is out! 🚀 github.com/predict-idlab/… Release notes ⬇️ - support serialization of downsamplers - improve & fix equidistant binning (~ searchsorted) - some (minor) performance improvements Downsampling 1B f32 data points in 0.06s ⚡️

🎉 Excited to introduce MinMaxLTTB, a time series downsampling algorithm co-developed with Jeroen Van Der Donckt, designed to boost the scalability of the LTTB algorithm 🚀! Our approach involves a two-step process: 1️⃣ Perform MinMax preselection 🎯 2️⃣ Apply LTTB on the preselected points

Pieter De Frenne European Research Council (ERC) GEMS is still alive 😍

Optioryx won the supply chain start-up award of the year award! 🏆🥂🍾🍻🥳