Differential privacy was mentioned a lot at #NeurIPS2019. Great to see open source projects from IBM Research with sklearn friendly syntax github.com/IBM/differenti… and a steady flow of additional models being supported.

Spurious correlations eye candy. Divorce rate vs margarine consumption is totally legit. tylervigen.com/spurious-corre…

Amazing effort!

powerful viz

Attending the virtual #widspugetsound2020 while working remotely and homeschooling is no joke. Great to see the efforts the organizers put into the networking! #wids2020

Stunning numbers: the lockdown prevented half a billion infections. This would mean at least 5M deaths, and probably towards 25M deaths because the hospital systems would be overwhelmed. That’s scale of deaths in WW II (~80M deaths). nature.com/articles/s4158…

"When workers come closer than six feet to one another, circles around their feet flash red on the TV, indicating to employees that they should move to a safe distance apart. "

Great check list for a Deep Learning pipeline.

Interested in co-hosting ML for Health #WiML breakout session at #icml2020 together with us (Tempest van Schaik PhD and Bianca Furtuna)? Let me know ASAP! WiML

Ashley Willis (McNamara)

Great news that the best paper at MIDL highlights the "false negative problem" in AI recon/denoising algorithms, whereby these models can *delete the features of pathology*, resulting in nice looking but non-diagnostic images. I wrote about this before: lukeoakdenrayner.wordpress.com/2019/10/23/the…

Join the discussion on "Healthcare and ML: real world applications and challenges" at WiML #icml2020 with Tempest van Schaik PhD , Bianca Furtuna , Katie Claveau and @SummerEla! Topics teaser: -Sensors & wearables, -ED triage for covid-19, -The TCR-antigen map project -Responsible AI!

Can’t wait to use Text Analytics for Health Cognitive Service — very impressive work!

Nice overview of approaches that can be used in-stream against IoT data — without the need ofhistorical data. These methods can range from simple techniques such as Lag monitoring to more complex ones like Subspace Tracking.

Primary care records of 17M adults were linked to 10,926 COVID-19-related deaths. These death were associated with: being male, older age and deprivation, diabetes, severe asthma. Compared with people with white ethnicity, Black and South Asian people were at higher risk...

Hiring! ;)

Great event and a line-up of presenters!

I wish someone showed me this earlier :)

totally