Weekly Review 30 July 2017

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

1. Adding 'imagination' to deep neural networks: http://www.theregister.co.uk/2017/07/21/deepmind_ai_imagination/
2. Half of low-skilled jobs will be replaced by AI / automation: http://www.techrepublic.com/article/50-of-low-skilled-jobs-will-be-replaced-by-ai-and-automation-report-claims/ 
3. What's old is new again in AI, we've just got better computers: http://www.theregister.co.uk/2017/07/21/artificial_intelligent/ 
4. How Google is using AI to help run its data centres: http://www.techrepublic.com/article/google-says-ai-will-help-run-datacenters-in-the-near-future/ 
5. Why quality, labelled data is so important for deep learning: https://techcrunch.com/2017/07/21/why-the-future-of-deep-learning-depends-on-finding-good-data/ 
6. The next version of HoloLens will have an onboard neural network chip: https://www.theverge.com/2017/7/24/16018558/microsoft-ai-coprocessor-hololens-hpu 
7. Overview of some of the hardware being applied to deep neural networks: http://www.theregister.co.uk/2017/07/24/ai_hardware_development_plans/ 
8. Qualcomm has open sourced its Neural Processing Engine: http://www.theregister.co.uk/2017/07/26/qualcomm_neural_processing_engine_freed/ 
9. A neural network based facial expression recogniser that gauges movie audience enjoyment: https://techcrunch.com/2017/07/25/this-facial-recognition-system-tracks-how-youre-enjoying-a-movie/ 
10. Diagnosing schizophrenia from brain scans using a support vector machine: http://www.theregister.co.uk/2017/07/26/ibm_and_uni_alberta_tackle_schizophrenia/ 
11. Elon Musk is a clever guy, but I think he is way off-beam about AI: https://www.theguardian.com/technology/2017/jul/25/elon-musk-mark-zuckerberg-artificial-intelligence-facebook-tesla 
12. AI is already causing problems, we should be worrying about them, not hypothetical killer robots: https://www.theverge.com/platform/amp/2017/7/25/16024444/ai-safety-threat-elon-musk-mark-zuckerberg 
13. Building decision trees in Scikit-Learn: http://www.dxbydt.com/decision-trees-classification-interpretation-using-scikit-learn/
14. Deep learning, AI, machine learning, made really simple: http://www.kdnuggets.com/2017/07/ai-deep-learning-explained-simply.html
15. Situations where you should not use deep learning: http://www.kdnuggets.com/2017/07/when-not-use-deep-learning.html 
16. List of Python machine learning tutorials: http://www.kdnuggets.com/2017/07/machine-learning-exercises-python-introductory-tutorial-series.html 
17. How to build a team that does AI: http://www.theregister.co.uk/2017/07/27/assembling_an_ai_team/ 
18. Nanoneurons for neuromorphic chips: http://spectrum.ieee.org/nanoclast/semiconductors/devices/nanoneurons-enable-neuromorphic-chips-for-voice-recognition
19. Current AI is so resource-intensive only a few big companies can effectively use it: https://www.datanami.com/2017/07/25/exposing-ais-1-problem/
20. Machines still struggle with conversational Chinese: https://www.technologyreview.com/s/608249/for-computers-too-its-hard-to-learn-to-speak-chinese/
21. Never mind terminators, this kind of AI can kill you just as dead: https://www.theverge.com/2017/7/28/16054834/mushroom-identifying-app-machine-vision-ai-dangerous