A song about the paper: Friedemann Zenke, Surya Ganguli. "SuperSpike: Supervised Learning in Multilayer Spiking Neural Networks", Neural Computation (2018) 30 (6): 1514–1541. [Intro] A rose beside the threshold, a thorn against the light. A thousand little almosts— and then, a spike. [Verse 1] You arrive in little flashes. I collect you as you come. But the room is always leaking; I forget the distant ones. Come a little closer, closer. There is something we could start. One more drop into the silence— then the sudden little spark. [Pre-chorus] Turn the dial a little—nothing. Turn it further—still the same. Then the whole room catches fire. One small turn became a flame. [Chorus] Keep the thorn. Let the rose lean toward the light. I can learn from almost even when I don’t ignite. Leave a little trace of it. Leave a little warmth. When the answer finds me, let it find me while the trace is warm. [Verse 2] There’s a cliff inside the question. There’s no gentle slope to climb. So I lay a trail of petals for the reaching hand to find. I still speak in sudden flashes. I still cross that narrow line. But the petals give a hint of where to reach next time. [Bridge — stripped back] You don’t have to be here when the message comes. First the meeting, then the memory— not a verdict, just a thread. Something missing asks for more. Too much asks for less instead. [Final chorus — layered vocals] Keep the thorn. Let the rose lean toward the light. I can learn from almost even when I don’t ignite. Leave a little trace of it. Leave a little warmth. When the answer finds me, let it find me while the trace is warm. [Outro] You were here. I was close. Something of the meeting stayed. A thousand little almosts. A slightly different next time.
