Saturday, May 28, 2016

Weekly Review 27 May 2016

Some interesting links that I Tweeted about in the last week:

  1. AI and the rise of the "useless" class:
  2. Overview of the applications of IBM's Watson:
  3. Machine learning is the "automation of automation":
  4. The promise of Google's AI:
  5. Google Home means holding conversations with computers - will they use those conversations to make better AI?
  6. A knowledge of measurement theory is really important: See also here: 
  7. China is really going in the wrong direction now. A real shame, I know lots of good people among China's academia:
  8. Australian robot livestock workers:
  9. Google is trying to teach an AI to be artistic:
  10. The (potential) contribution of AI to medicine:
  11. Some thoughts on language choice for writing a web crawler: Last crawler I wrote was in Python.
  12. Natural language processing and AI in Facebook:
  13. Facebook is planning on using neural networks for translation: That'll need some really, really big neural networks.
  14. What's good and what's bad about TensorFlow:
  15. The disappointment of AI personalisation:
  16. More than just bots in the intelligent application ecosystem:
  17. Machine learning algorithms that learn from fewer examples:
  18. Biased data will give you biased models: I used to teach this to my third-year AI class, why don't pros know?
  19. Why Facebook's AI can't recognise a mirror selfie:
  20. Terrapattern is a reverse image searching on maps, powered by a convolutional neural network:
  21. A biased data set will give a biased model, even if the biases are racial/gender/cultural. Why is this still news?
  22. At least the US government is taking AI seriously. Will others?
  23. Why Python is such a good match for machine learning: 
  24. Something of a glossary of key machine learning terms:
  25. Amazon is expanding its cloud-based machine learning offerings: