One of the factors explaining the power of a nervous system is the fact that computation takes place within the very entities where information is stored.
This has led the electronics industry to explore, for years, the possibility of building electronic neurons. But the emulation of neuron behavior is far from obvious and the electronic neurons that we’ve built so far are big, are made out of multiple elements (transistors and capacitors) and consume too much energy.
But last year engineers Mario Lanza and Sebastian Pazos, at Singapore University have made a major breakthrough in this field: they’ve replicated the neuron behavior using a single transistor. This opens a path for low-power AI solving in part the energy issue associated with data centers.
To illustrate the importance of this finding, we’ve made a video covering the history of the transistor and explaining how Lanza and Pazos have made this compact electronic neuron.
There is also a Spanish version of the video: