- #1
- 2,810
- 604
After watching "Still Alice"(starring Juliane Moore) which is about an Alzheimer's patient, I started thinking about this disease a little bit and suddenly it occurred to me that its plausible to expect seeing such an effect in neural networks too. I mean, imagine you have a neural network that solves e.g. Schrodinger's equation. As time passes and it solves more equations, it should become better at it. But suddenly you see that its ability is decreasing gradualy as it solves more equations which should seem strange.
Does anyone know about an observation of such an effect in a neural network? Is there any theoretical prediction of such an effect? If yes, what's the minimum complexity needed for such an effect to happen?
Thanks
Does anyone know about an observation of such an effect in a neural network? Is there any theoretical prediction of such an effect? If yes, what's the minimum complexity needed for such an effect to happen?
Thanks