Friday, November 12, 2021

Weekly Review - 12 November 2021

Some interesting links that I Tweeted about this week:

1) Simulating a dragonfly's brain shows that enormous deep neural networks aren't always necessary: https://spectrum.ieee.org/fast-efficient-neural-networks-copy-dragonfly-brains

2) The precariousness of research positions is why I got out of a research-only career and into academic management - a much better way of supporting my family: https://www.nature.com/articles/d41586-021-02613-4

3) Some recent advances in reinforcement learning for deep  neural networks: https://spectrum.ieee.org/reinforcement-learning

4) AI in general, and Deep Neural Networks in particular, have been over-hyped. They can still achieve some useful results, but a healthy level skepticism should be employed: https://www.bloomberg.com/opinion/articles/2021-10-04/artificial-intelligence-ain-t-that-smart-look-at-tesla-facebook-healthcare


6) I've been working from home almost three months. I've never done any of these five things: https://www.stuff.co.nz/business/prosper/advice/300419444/five-things-to-avoid-when-working-from-home

7) This is not going to end well: https://spectrum.ieee.org/predictive-ai-pentagon AI


9) Organic, implantable, neural networks https://spectrum.ieee.org/brain-inspired-ai-implant 

10) My alma mater is using AI to increase energy efficiency in its buildings: https://www.odt.co.nz/news/dunedin/campus/ai-managing-university-energy-use

11) Deep learning in military robots: https://spectrum.ieee.org/ai-army-robots

12) Good overview of time-series prediction with machine learning: https://www.kdnuggets.com/2020/03/machine-learning-time-series-forecasting-sequel.html

13) How the energy costs of deep learning are becoming unsustainable: https://spectrum.ieee.org/deep-learning-computational-cost


15) Overview of the history of AI, especially neural networks: https://spectrum.ieee.org/history-of-ai

No comments:

Post a Comment