Tuesday, May 23, 2017

Weekly Review 23 May 2017

Below are some of the interesting links I Tweeted about recently.

  1. The importance of visual (image) data to future developments in AI: https://techcrunch.com/2017/05/17/the-war-over-artificial-intelligence-will-be-won-with-visual-data/
  2. One-shot visual learning of tasks from humans in VR with neural networks: http://www.theregister.co.uk/2017/05/16/openai_teaches_robot_to_learn_from_human_demonstrations_in_vr/ 
  3. Detecting masks and guns in CCTV feeds, using deep learning: https://techcrunch.com/2017/05/16/deep-science-ai-monitors-security-feeds-for-masks-and-guns-to-quicken-response-times/ 
  4. Machine learning can eliminate bias in hiring, but only if the training data is unbiased: http://www.techrepublic.com/article/how-machine-learning-can-help-companies-eliminate-bias-in-hiring/ 
  5. Diagnosing concussions with deep learning: https://techcrunch.com/2017/05/16/brightlamp-wants-to-use-ai-to-spot-concussions/ 
  6. Conference submission deadline: ICONIP 2017 https://computational-intelligence.blogspot.com/2017/05/conference-submission-deadline-iconip.html 
  7. How to build a text classifier utilising machine learning: https://blog.monkeylearn.com/how-to-create-text-classifiers-machine-learning/
  8. This guy does not like the R language, and not without some good reasons: http://www.datasciencecentral.com/profiles/blogs/why-r-is-bad-for-you 
  9. Academic resumes tend to be long, but it helps to put the main points on the first or second page: https://www.seek.co.nz/career-advice/5-things-employers-wish-they-could-say-about-your-resume
  10. 5 different types of recommenders explained: http://www.datasciencecentral.com/profiles/blogs/5-types-of-recommenders 
  11. Google has brought TensorFlow to Android: https://techcrunch.com/2017/05/17/googles-tensorflow-lite-brings-machine-learning-to-android-devices/
  12. A neural network that invents new paint colours and names: http://lewisandquark.tumblr.com/post/160776374467/new-paint-colors-invented-by-neural-network "Clardic fug" is probably my favourite paint name.
  13. How to build, and convert into code, decisions trees in Python: http://www.kdnuggets.com/2017/05/simplifying-decision-tree-interpretation-decision-rules-python.html 
  14. Google's AI is starting to manage relationships between people: https://www.theverge.com/2017/5/19/15660610/google-photos-ai-relationship-emotional-labor 
  15. AI in agriculture-a bit vague about what the AI is, though: http://www.datasciencecentral.com/profiles/blogs/artificial-intelligence-in-agriculture-what-s-next 
  16. Google is making its Tensor Processing Units available via the cloud: https://www.datanami.com/2017/05/18/cloud-tpu-bolsters-googles-ai-first-strategy/ 
  17. An argument that AI is not driving cloud adoption: https://www.theregister.co.uk/2017/05/18/baby_steps_not_ai_drives_cloud_growth/ 
  18. So Canada's developing a nation-wide AI strategy? http://www.datasciencecentral.com/profiles/blogs/pan-canadian-artificial-intelligence-strategy 
  19. An AI that is "outrageous in a great way": http://www.theregister.co.uk/2017/05/22/deeptingle_ai_transforms_writing/