Sunflower
09/20/2024 (Fri) 00:51
Id: 7be093
No.9089
del
>>9087
I "get it" conceptually, but not how the weights are updated. It's like the police policing itself, you know?
Making it very simple the weights are thresholds that would for example make a star shaped object end up in slot 1, a round object in slot 2 and a square in slot 3, by the thresholds having a special shape that lets through exactly those shapes. But why would they change to do this? That's where you'd used reinforcement or monitored training, but it's supposed to represent a brain. You let data flow from one end to the other, then give feedback which says "right" or "wrong" which then changes the weights. This feedback would then access the setting on the weights in relation to that specific object. How?
I "get it" conceptually, but not how the weights are updated. It's like the police policing itself, you know?
Making it very simple the weights are thresholds that would for example make a star shaped object end up in slot 1, a round object in slot 2 and a square in slot 3, by the thresholds having a special shape that lets through exactly those shapes. But why would they change to do this? That's where you'd used reinforcement or monitored training, but it's supposed to represent a brain. You let data flow from one end to the other, then give feedback which says "right" or "wrong" which then changes the weights. This feedback would then access the setting on the weights in relation to that specific object. How?