', 'auto', 'clientTracker'); ga('clientTracker.send', 'pageview');
Sep 14, 2020
Gordon Wilson is the co-founder and CEO of Rain Neuromorphics. They work on technologies that improve memory processing, analog computation, and scalable systems. In this episode, Gordon Wilson discusses how Rain Neuromorphics recreates neural networks.
Top three takeaways:
[0:00] Ladan introduces the episode and the guest, Gordon Wilson.
[2:40] Wilson describes why building a brain is necessary for understanding it.
[4:40] In this research, it is fundamental to understand the difference between digital and analog signal processing.
[7:10] Processing in memory consists of using analog processors to complete matrix math.
[9:40] The Rain Neuromorphics technology allows scaling up of analog processing to build larger neural networks.
[12:30] Current projects include fabricating nano-wires that perform matrix multiplication.
[14:50] Chips that are analog and scalable are very well suited to be the kind of device in brain processing.
[16:30] The “memristor” is programmable and capable of changing resistance values.
[19:30] Rain Neuromorphics completes multidisciplinary projects in the Bay area.
[21:47] The product that Rain Neuromorphics will bring to the market will be a massive, sparsely connected array of neurons.
[24:40] A “puff” occurs when wires are placed in a stochastic fashion in such a way that they are physically unclonable.