Weekly Review 29 April 2022

Some interesting links that I Tweeted about this week:

1) Kids do get AI, and technology in general. My own kid engages with AI very easily. https://spectrum.ieee.org/kids-ai

2) Not DevOps, but MLOps: https://www.kdnuggets.com/2021/11/accelerating-ai-mlops.html

3) Evolutionary robotics. Not a new idea, but the simplification of it to two dimensions speeds things up a bit: https://spectrum.ieee.org/robot-design

4) Reproducibility has been a problem in science for a while now. How much #AI research is repeatable? https://www.nature.com/articles/d41586-021-02486-7

5) Moving #NeuralNetwork operations to analog devices: https://spectrum.ieee.org/new-devices-for-analog-ai

6) There is a lot of hype around #AI, but tech experts seems to be less vulnerable to it: https://www.techrepublic.com/article/our-most-technical-people-are-down-on-ai-and-thats-a-good-thing/

7) Stephen Grossberg is not a fan of #DeepLearning, and prefers his own ART model: https://spectrum.ieee.org/deep-learning-cant-be-trusted

8) Quantum entanglement can boost the performance of quantum #MachineLearning: https://spectrum.ieee.org/quantum-machine-learning

9) Overview of transfer learning in #AI: https://www.kdnuggets.com/2022/01/transfer-learning.html

10) So now an #AI can be recognised as an inventor for a patent? This will not end well: https://spectrum.ieee.org/first-time-ai-named-inventor

11) Using #MachineLearning to detect depression from EEG signals: https://www.odt.co.nz/news/dunedin/using-ai-help-treat-depression

12) The six worst case scenarios that could be brought about by #AI. None of which are extermination of the human race: https://spectrum.ieee.org/ai-worst-case-scenarios

13) An overview of concept drift in #MachineLearning. This is what causes the performance of deployed #AI models to degrade over time: https://www.kdnuggets.com/2022/01/machine-learning-models-die-silence.html

14) Using #DeepLearning to detect early Alzheimer's from speech samples: https://spectrum.ieee.org/ai-to-detect-alzheimers

15) So medical imagers "hallucinate" because they use #MachineLearning to process the images before the operator sees them? I really don't think that's a good idea: https://spectrum.ieee.org/ai-medical-imaging-false-structures

16) Nine technologies that will help with your advanced studies in #MachineLearning. Wish I'd had some of these when doing my PhD: https://www.kdnuggets.com/2021/09/nine-tools-mastered-before-phd-machine-learning.html

17) Building a synthetic world to help create training data for #AI: https://spectrum.ieee.org/synthetic-data-ai

18) Diagnosing prostate cancer biopsies with #AI: https://www.odt.co.nz/news/dunedin/campus/ai-can-diagnose-prostate-cancer-study

19) Faster #NeuralNetwork processing with wafer chips: https://spectrum.ieee.org/graphcore-ai-processor

20) I think most researchers have felt burnt out, or felt like an imposter, at some point in their careers: https://www.nature.com/articles/d41586-021-03042-z