[ 3 / biz / cgl / ck / diy / fa / ic / jp / lit / sci / vr / vt ] [ index / top / reports ] [ become a patron ] [ status ]
2023-11: Warosu is now out of extended maintenance.

/sci/ - Science & Math

View post   
File: 219 KB, 990x698, 1423479254530.png [View same] [iqdb] [saucenao] [google]
8456103 No.8456103 [Reply] [Original]

Psychological issues are equivalent to incorrectly trained neural nets which produce less than optimal results for some inputs. Psychological issues are movements away from a state of true free will

Is there such thing as a self repairing neural net data structure which can detect over or underfitting and rectify it in a general mathematical way? And not nocessarily become optimal but become better on average over time?

This is equivalent to the problem of free will

>> No.8456135

>>8456103
'incorrectly trained neural nets' is silly. some psychological issues are caused by diseases at the neuron level; others are high level biases caused by your education and linguistic framework. not to mention we don't understand the graph even a bit. so any assumption like that is completely baseless.

>> No.8456149

>>8456135
Education and linquistic framework is the training data.

>> No.8456160

The brain is stratified with fractal feedback loops on each and every level. These correspond to deep learning with application such as long short term memory where the epsilon distance is the trainable thresold. Your idea is not new. Also Godel Escher Back general.