Profile cover photo
Profile photo
Michael McGreevy
Michael's interests
View all
Michael's posts

Post has shared content
This has some pretty cool privacy properties.
Standard machine learning approaches require centralizing the training data on one machine or in a datacenter. And Google has built one of the most secure and robust cloud infrastructures for processing this data to make our services better. Now for models trained from user interaction with mobile devices, we're introducing an additional approach: Federated Learning. 

Post has attachment

Post has attachment
"I don't want to denigrate Kazakhstan, or indeed their artistic skills with movies like Borat," Senator Birmingham said.

Ladies and Gentlemen, our education minister.

Post has attachment
This is so so awesome.

Post has attachment
Michael Caine is Michael Caine. Not a lot of people know that 😀. This article made me think of .

Post has attachment
It looks like the ABC lost the keys to the computer they use to update the "Latest Top Photos" section of their mobile news site. It's been showing the same photos for over two months now:

Post has attachment

Post has shared content
It's not all about moonshots.
Google's re:work team has published in their blog a one-pager I wrote internally at Google a couple of years ago about one of our most successful (and less publicized) approaches to innovation.

Post has shared content
A good cause.
My friend Monique is probably one of the bravest people I know. Not only did she face the loss of her son to cancer with great courage, but she also went on shaving her hair all off yesterday morning to raise funds and awareness to fight this horrible disease.

Please show her some support. To make a donation,

To create awareness, please share this post.

Thank you.

Post has shared content
So awesome.
Tensor Processing Units (TPUs)
I'm very excited that we can finally discuss this in public. Today at Google I/O +Sundar Pichai revealed the TPU (Tensor Processing Unit), a custom ASIC that Google has designed and built specifically for machine learning applications. We've had TPUs deployed in Google datacenters for more than a year, and they are an order of magnitude faster and more power efficient per operation than other computational solutions for the kinds of models we are deploying to improve our products. This computational speed allows us to use larger, more powerful machine learned models, expressed and seemlessly deployed using TensorFlow ( into our products, and to deliver the excellent results from those models in less time.

TPUs are used on every Google Search to power RankBrain (, they were a key secret ingredient in the recent AlphaGo match against Lee Sedol, they are used for speech and image recognition, and they are powering a growing list of other smart products and features.

+Norm Jouppi and the rest of the team that developed this ASIC did a fabulous job, and it's great to see it discussed in public!

Blog post:

Link to the part of the keynote where Sundar discusses TPUs:

WSJ article:

Edit: Added a link and some text.
Wait while more posts are being loaded