Linkstream – Telegram
Linkstream
173 subscribers
32 photos
3 videos
2 files
899 links
Various links I find interesting. Mostly hardcore tech :) // by @oleksandr_now. See @notatky for the personal stuff
Download Telegram
Google support at its best
(TLDR: Firebase costs increased 70x because they started including SSL handshake in traffic calculation)
https://news.ycombinator.com/item?id=14356409
On the Turing Completeness of MS PowerPoint 😂
https://www.andrew.cmu.edu/user/twildenh/PowerPointTM/Paper.pdf
Neurosurgeon: Collaborative Intelligence
Between the Cloud and Mobile Edge
TLDR: It's best to split processing between device and cloud, here's how to do it automatically (for DNNs)
http://web.eecs.umich.edu/~jahausw/publications/kang2017neurosurgeon.pdf
DeepXplore: Automated Whitebox Testing of Deep Learning Systems '17
TLDR: Neuron coverage, use 3+ cross-referencing oracles to detect outliers
https://arxiv.org/pdf/1705.06640.pdf
> Design your construct, pick your deliverable - dsDNA, plasmid DNA or purified protein - and Serotiny will instantly price your order from various vendors.
https://serotiny.bio/notes/pinecone/
We live in interesting times :)
> Big data systems may scale well, but this can often be just because they introduce a lot of overhead
I believe I already posted this one, but somehow I can't find it in the history, so here it comes again
http://www.frankmcsherry.org/graph/scalability/cost/2015/01/15/COST.html
CNTK 2.0. Nice production-related features, note Model Compression, Keras support and impressive multi-GPU scaling.

> If your project involves sequence processing, such as speech, natural language understanding, machine translation, etc., CNTK should be your best choice. And if you are a vision researcher working on video processing, you should definitely give CNTK a try.

https://www.microsoft.com/en-us/cognitive-toolkit/blog/2017/06/microsofts-high-performance-open-source-deep-learning-toolkit-now-generally-available/
and now for something completely different
http://4dtoys.com/
Unicode nuances you probably don't need to know about :)
http://www.unicode.org/reports/tr29/#Grapheme_Cluster_Boundaries
Exploring LSTMs - with pretty pictures
http://blog.echen.me/2017/05/30/exploring-lstms/