Saturday, May 7, 2016

Weekly Review 6 May 2016

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

  1. Google DeepMind has been given access to records on 1.6M British patients: http://www.theregister.co.uk/2016/04/29/google_given_access_to_reams_of_confidential_patient_information/
  2. Mark Zuckerberg says machine learning will exceed human performance in speech & image recognition in 5-10 years: http://www.datanami.com/2016/04/29/ai-surpass-human-perception-5-10-years-zuckerberg-says/
  3. Do deep learning image processing systems really see like humans do? http://motherboard.vice.com/en_au/read/computers-might-just-see-like-humans-after-all-vision-deep-learning-neural-networks
  4. Top ten programming languages for (retaining) employment: http://www.informationweek.com/devops/programming-languages/10-programming-languages-that-will-keep-you-employed-/d/d-id/1325314 
  5. Identifying heart attacks with AI: http://www.huffingtonpost.com/abhinav-sharma/can-artificial-intelligen_2_b_9798328.html
  6. Machine learning at Facebook-goal of one AI agent per user: http://fortune.com/facebook-machine-learning/
  7. Princeton's professors CV of failure: http://www.theguardian.com/education/2016/apr/30/cv-of-failures-princeton-professor-publishes-resume-of-his-career-lows The lesson here seems to be persistence is important, but so is luck.
  8. The arguments in favour of journal paywalls just don't hold water: http://www.slate.com/articles/health_and_science/science/2016/04/science_magazine_can_t_defend_its_flawed_business_model.html
  9. Why business intelligence tools need statistics (and actual intelligence): http://www.theregister.co.uk/2016/05/02/stats_the_problem_with_bi/
  10. Qualcomm is releasing a SDK for its deep learning platform: http://www.theverge.com/2016/5/2/11538122/qualcomm-deep-learning-sdk-zeroth
  11. Networking for data scientists: http://www.kdnuggets.com/2016/05/how-network-build-personal-brand-data-science.html Same principles apply to networking for computational intelligence researchers
  12. AI in financial trading: http://www.euromoneythoughtleadership.com/ghostsinthemachine/
  13. Infosys has developed a system using AI for knowledge management: http://techcrunch.com/2016/04/28/new-infosys-ai-tool-could-change-the-way-companies-maintain-complex-systems/
  14. How long to true AI? http://dataconomy.com/far-away-inventing-true/ Time to re-read my copy of Kurzweil's The Age of Intelligent Machines
  15. Recent surveys of evolutionary algorithms: http://cis.ieee.org/index.php?option=com_content&view=article&id=557:the-latest-surveys-of-evolutionary-algorithms-updated-3&catid=17:e-newsletter-news-a-announcements
  16. Is Uber planning to use machine learning to eliminate surge pricing? http://www.npr.org/sections/alltechconsidered/2016/05/03/476513775/uber-plans-to-kill-surge-pricing-though-drivers-say-it-makes-job-worth-it
  17. An overview http://techemergence.com/elon-musks-ai-gym-lets-you-develop-and-train-ai-algorithms/ of the Open "AI Gym" https://gym.openai.com/ for testing reinforcement learning algorithms.
  18. The US White House is running workshops to consider the future implications of AI: https://www.whitehouse.gov/blog/2016/05/03/preparing-future-artificial-intelligence 
  19. Quantizing deep learning neural networks in TensorFlow: http://www.kdnuggets.com/2016/05/how-quantize-neural-networks-tensorflow.html
  20. Learning Python for data science: http://www.datasciencecentral.com/profiles/blogs/the-guide-to-learning-python-for-data-science-2
  21. Machine learning and big data in credit card companies: http://www.datanami.com/2016/05/03/credit-card-companies-evolving-big-data/
  22. This. This is why I will not review any paper for any Elsevier journal. They will not profit from my free labour: https://torrentfreak.com/elsevier-complaint-shuts-down-sci-hub-domain-name-160504/
  23. AI in medicine is coming: http://m.nzherald.co.nz/technology/news/article.cfm?c_id=5&objectid=11634284
  24. Art critics do not like art produced by AI: https://www.theguardian.com/science/2016/may/06/does-an-ai-need-to-make-love-to-rembrandts-girlfriend-to-make-art
  25. Can machine learning make us more beautiful? http://dataconomy.com/better-botox-beauty-industry-get-ai-makeover/
  26. Apparently AI is still too stupid to pose an existential threat to humans: http://www.techrepublic.com/article/microsoft-research-chief-ai-is-still-too-stupid-to-wipe-us-out-and-will-be-for-decades/
  27. User behaviour analysis using AI is a growing trend in cyber-security: http://www.darkreading.com/threat-intelligence/silicon-and-artificial-intelligence-the-foundation-of-next-gen-data-security/a/d-id/1325401
  28. I put more stock in Adam Coates'  opinions on the dangers of AI than Stephen Hawking's: http://www.informationweek.com/big-data/big-data-analytics/artificial-intelligence-dont-fear-it-embrace-it/d/d-id/1325391
  29. Viv the Siri killer: https://www.technologyreview.com/s/601401/siri-killer-viv-faces-an-uphill-battle/
  30. Researchers open to other disciplines produce better research: https://www.insidehighered.com/news/2016/05/06/new-paper-suggests-open-minded-researchers-produce-higher-quality-research Working with ecologists made me a better researcher.
  31. AI in staff recruitment: http://www.techrepublic.com/article/how-ai-can-help-companies-hire-and-retain-the-best-employees/
  32. Are postdocs actually harmful to researchers' careers? https://www.insidehighered.com/news/2016/05/06/study-graduate-students-may-take-unnecessary-postdocs 
  33. How to improve customer experience with AI: http://www.cmswire.com/customer-experience/3-ways-to-bring-artificial-intelligence-to-your-customer-experience/
  34. Teaching deep learning ANN to be more "conversational"a by training them on romance novels: https://www.buzzfeed.com/alexkantrowitz/googles-artificial-intelligence-engine-reads-romance-novels?utm_term=.hyZr7QWEl#.dfPM42YKG
  35. An argument that computers do not either learn nor predict: http://techcrunch.com/2016/05/05/only-humans-not-computers-can-learn-or-predict/