3 ms·
> Analog computers making a comeback ? you never know. For dataflow paradigms like deep neural nets it makes sense. If it's significantly faster and significan
by breck 7y ago
> Analog computers making a comeback ? you never know.
For dataflow paradigms like deep neural nets it makes sense. If it's significantly faster and significantly more power efficient, and the intermediate states are largely arbitrary weights anyway, what's the point of digital precision.
- regularfry 7y agoIt's not the precision, it's avoiding O(n^2) physical connections per layer.
- breck 7y agoThat's my point. Swapping digital for analog gives up the ability to know the precise bits of a register but in return you gain a lot in efficiency.
- regularfry 7y agoThat's not what I'm getting at. It's not the nodes you need to worry about, it's the edges. If you've got two fully connected 100-neuron layers in a physical circuit, that's 10,000 wires you need to route, and they all want to cross each other. That's why nobody builds physical neural networks.