Vitaly retweeted
Entropy is not the easiest thing to understand.
It is rumored to describe something about information and disorder, but it is not clear why.
What do logarithms and sums have to do with information?
💡 Let me explain! 💡
Vitaly retweeted
Happy to announce telesto.ai, our project to make machine learning crowdsourcing production-ready! We are releasing our platform early and would like to continue the development with community feedback! (1/8)
Vitaly retweeted
Being resource-constrained forces you to ship solutions that do less, or do it in a simpler way. Often, these solutions turn out to be more effective than the over-engineered solutions shipped by resource-rich teams.
Resource constraints can be a design advantage.
Vitaly retweeted
Tip: excalidraw.com is fantastic for quickly mocking up UIs. Free, feature-filled and the live collaboration mode is excellent.
Great work by @Vjeux and team!
Vitaly retweeted
We've made a draft of our deep learning book available as @ProjectJupyter notebooks.
Important: This is for personal use, and not licensed for any redistribution or change of format or medium. Eg. don't make a PDF and share it, please!
github.com/fastai/fastbook
Vitaly retweeted
Glad to see Catalyst success in both @kaggle and @NeurIPSConf – 4th place on Recursion Cellular Image Classification!
When it comes to reproducibility, teamwork and fast R&D – Catalyst is a clue.
writeup: kaggle.com/c/recursion-cellu…
code: github.com/ngxbac/Kaggle-Rec…
Vitaly retweeted
Our paper on training with pseudo-labels for semantic segmentation, GCPR 2019.
Semi-Supervised Segmentation of Salt Bodies in Seismic Images:
SOTA (1st place) at TGS Salt Identification Challenge.
🌐 github.com/ybabakhin/kaggle_…
📝 arxiv.org/abs/1904.04445
#kaggle #TGS2019 #GCPR19
Vitaly retweeted
Look how folks at @getqonto are using DVC to version inputs, intermediate files and algorithm models💪
Reading that DVC is helping to 'drastically increase productivity' is making us so happy!🎯👏
medium.com/qonto-engineering…
Vitaly retweeted
#AI and #deeplearning people! So, recently I implemented a #CNN from scratch, essentially building a mini DL framework. Would anyone be interested if I were to expand how to build a general DL framework with modular components as a series of Jupyter notebook lessons?
Vitaly retweeted
Catalyst 19.06 has been released!
Lots of improvements:
- distributed training with fp16 support
- updated segmentation models contrib
- on-policy reinforcement learning
Read the release notes here ↓
github.com/catalyst-team/cat…
Vitaly retweeted
'Tips and Tricks for Machine Learning' a live presentation from #KaggleDays Paris by Kaggle Grandmaster Stanislav Semenov. [WATCH] ow.ly/nqlq30ocxIz // @Kaggle_Days
Vitaly retweeted
Have you ever found your expert annotated microscopy dataset too small? We did too, that's why we developed a style transfer based method to boost the training dataset by super realistic fake images! Check it out! biorxiv.org/content/10.1101/…
#deeplearning #microscopy #singlecell
Vitaly retweeted
In collaboration with Google, we're releasing Activation Atlases: a new technique for visualizing what interactions between neurons can represent.
💻Blog: blog.openai.com/introducing-…
📝Paper: distill.pub/2019/activation-…
🔤Code: github.com/tensorflow/lucid/…
🗺️Demo: distill.pub/2019/activation-…
Vitaly retweeted
Seeking for a #fullstack #software #engineer to join our R&D @alexandrovteam @embl Heidelberg to work on @metaspace2020, @cloudbutton2020, other projects! #science #Python #Javascript #opensource #cloud #computing #serverless embl.de/jobs/searchjobs/inde… Plz RT!
Vitaly retweeted
Dennis Graham has started a new career at age 73 in deep learning, and is doing cutting-edge work on diagnosing Parkinson's disease. fast.ai/2019/02/21/dl-projec…
ALT At age 73, Dennis Graham is using deep learning to diagnose Parkinson’s disease from Magneto-Encepholo-Graphy (MEG), as part of a UCH-Anschutz Neurology Research center project. Dennis is painfully familiar with Parkinson’s, as his wife has been afflicted with it for the last 25 years. MEG has the advantages of being inexpensive, readily available, and non-intrusive, but previous techniques had not been analytically
Vitaly retweeted
We've trained an unsupervised language model that can generate coherent paragraphs and perform rudimentary reading comprehension, machine translation, question answering, and summarization — all without task-specific training: blog.openai.com/better-langu…
Vitaly retweeted
This is a super cool resource: Papers With Code now includes 950+ ML tasks, 500+ evaluation tables (including SOTA results) and 8500+ papers with code. Probably the largest collection of NLP tasks I've seen including 140+ tasks and 100 datasets.
paperswithcode.com/sota
Vitaly retweeted
Practical Deep Learning for Coders, 2019 edition, is now available. With a shiny new video player with searchable transcripts.
This course is 100% new material, including some new techniques and results never previously published.
fast.ai/2019/01/24/course-v3…
