Google   530532

« earlier    

Google for Education: Computer Science Research
Find all of Google’s Computer Science Education resources in one place. For resources, programs, and funding opportunities visit
google  computerscience  educationresearch 
yesterday by actionhero
We have broken SHA-1 in practice.
cryptography  google  security  collision  sha-1  broken 
yesterday by gdw
DeepMind in talks with National Grid to reduce UK energy use by 10% | Ars Technica UK
AlphaGo merely en passant. Now DeepMind wants to cut power usage and global warming.
deepmind  google  ai  deeplearning  ramblerai 
yesterday by pklemenkov
Google AI detects breast cancer better than pathologists • Pharmaphorum
Marco Ricci:
<p>After ‘training’ the algorithm, researchers were able to achieve a 92% sensitivity in picking out tumour cells from the slides – significantly higher than the 73% achieved by trained pathologists with no time constraint.

In addition, the team recreated the accuracy in different datasets taken from other hospitals and scanning machinery.

The team did report an average of eight false positive per slide compared to none from trained pathologists. However, this rate was lowered through further customisation of the algorithm.

The algorithm results display as a heat map (shown below) that overlay a particular colour in regards to the likelihood of a specific part of an image housing cancerous cells.

<img src="" width="100%" />
<em>Left: Pathology slides of two lymph node biopsies. Middle: early results from deep learning tumour detection. Right: final results following further customisation of the algorithm.</em>

In everyday practice, pathology slide analysis is a time-consuming process, especially when considering that each patient typically has 10 or more pathology slides for their suspected tumour. Even when the pathology process is complete, it does not always lead to a definitive diagnosis.

Because of this, using machines to accurately analyse imagery for disease diagnosis, providing a definitive decision in a much shorter time period, is an area of particular excitement in the ongoing adoption of AI in healthcare.</p>

There must be a complicated calculation to make about how few false positives and negatives you hit before you let the machine do all the diagnosis and the doctors get on with more important stuff, such as talking to the patients.
cancer  google  diagnosis 
yesterday by charlesarthur
Google Maps will now let you share your location, creating a whole new set of privacy concerns • Recode
Tess Townsend:
<p>the updates don’t mark a sweeping change as the company has been careful about how it tweaks the service. That’s because Maps is Google’s most-used app after YouTube and the fourth most used app overall with <a href="">over 95 million people accessing it every month</a>, according to comScore. Maps has become crucial to Google’s mobile strategy.

Given that, it’s noteworthy that the changes don’t include any new ways for Google to make money from Maps.

Location sharing is the most significant update. People can let anyone else know where they are by sending a text message with a link. The link can be opened by anyone, even if they don’t have the Maps app.

That could raise all kinds of privacy concerns. The links, for example, can be shared on to anyone else through a simple copy and paste, whether or not the original user intended their information to be known to a wider circle.

But Maps product manager Ben Greenwood noted there are already ways a person could share your location. “It’s also possible they could take a screenshot of where you are,” he added.

Additionally, users will be able to share their location for set periods of time, or indefinitely until they turn the feature off. People will receive email reminders every two or three weeks that it’s still on.

That level of location sharing could be a problem in an abusive relationship where one person could demand the other keep the feature turned on, making it easier for them to track their whereabouts.</p>

This sort of location-sharing has been in Apple's "Find My Friends" app for ages (in my family, we call it "Stalk My Friends"). Except you can't copy-paste a link; you have to give permission explicitly to let someone track you.

Another thought: if there are 95m Google Maps users in the US, and Apple Maps (60m users, in the linked ComScore slide deck) are used 3.5x more than Google Maps on iOS, that means 17m Google Maps users on iOS, and so 78m Google Maps on Android users. If no Google-Maps-on-iOS user uses Apple Maps, that means 77m iPhone users, and 78m Android users in the US. But that seems too close; there's surely overlap on the Apple side.
google  apple  maps 
yesterday by charlesarthur

« earlier    

related tags

#tw  %google_maps  2017-03-20  2017-03-21  2017-03-22  2017-03-23  2017  360video  a.i.  accelerated_mobile_pages  advertising  ai  algorithm  amazon  amp  analysis  analytics  android  api  apple  ar  architecture  art  article  auth  automation  av  backup  barons  big  bigquery  blogroll  branding  breakthrough  broken  bucket  business  business_strategy  calendar  call  cancer  cap  car  chrome  climate-change  cloud  code  coding  collision  comp3911  compression  compute  computerscience  conference  content  cost-benefit  cryptography  daring_fireball  data  debug  deeplearning  deepmind  demo  design  devops  diagnosis  distributed-systems  dns  drawing  ebook  economics  edina  educationresearch  email  encoder  encoding  energy-resources  engineering  englisch  english  environment  facebook  fakenews  finance  find  firebase  font  for  foursquare  fraud  free_speech  fronteers@netvlies  gae  gcp  gcs  generation  github  gmail  gold  google  googlemaps  gps  graphics  guetzli  guides  hackernews  hate  helps  here's  hosting  how  howtos  identity  image  image_recognition  images  imagetools  inbox  incentives  infrastructure  inspiration  insurance  internet  internets  journalism  jpeg  jpg  json  latitude  launcher  law  legal  lib  library  linguistics  links  location-based_apps  location  logo  longitude  mac  machine_learning  machinelearning  manufacturing  maps  marketing  merge  mike_bostock  millennial  money  news  now  online  org:sci  osx  outlook  pagespeed  pairings  parenting  parents?  pay  performance  phishing  photo  platform  play  player  policing  politics  presentation  privacy  programming  protobuf  protocols  publishers  publishing  pwa  python  quality  ramblerai  redefining  research  reverse_engineering  s3  search  security  seo  serialization  sha-1  size  sm  social.messaging  software  sphericalvideo  spreadsheet  sre  stackoverflow  startups  storage  streetview  surveillance  tableau  technologia  technology  tema  template  testing  their  they're  think  tips  to  tool  tools  totwitter  training  unicode  usa  video  videos  visual  visualization  visualsearch  voice  voiceinterface  vps  vr  wallpaper  warrant  web  webinar  webinars  wifi  with  xss  your  youtube   

Copy this bookmark: