Together with the arrival of the AI we’ve understood that its mammoth energy demand is going to drive to a crisis. And that’s why a lot of scientists are focusing on producing better ways to compute AI-like algorithms.
In this new paper, with contributions of institutions from China, the UK, Spain or Uzbekistan, scientists open the path that would allow the use of skirmions and their spin textures, for energy-efficient neuromorphic systems. Preliminary results show decreases in energy consumption of several orders of magnitude, while been highly tunable and reaching recognition accuracy of 96% in handwritten digit recognition.
This picture, design in close collaboration with Prof. Elton Santos, has been featured on the cover of Advanced Materials.