Timeline for Computational Power of Neural Networks?
Current License: CC BY-SA 2.5
7 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Nov 4, 2010 at 18:33 | comment | added | Yaroslav Bulatov | A note on terminology -- important information is how many hidden layers there are. Zero hidden layer neural network with one output is just a linear threshold function, and is often (confusingly) called one layer or two layer neural network/perceptron, depending on whether inputs/outputs are considered layers. Also, in AI literature, neural networks are typically defined in terms of sigmoid functions which means that input/outputs are real valued. One hidden layer networks are known to be universal approximators in a sense that any continuous function can be approximated arbitrarily close | |
| Nov 4, 2010 at 16:41 | history | edited | András Salamon |
edited tags
|
|
| Nov 4, 2010 at 5:32 | vote | accept | gabgoh | ||
| Nov 4, 2010 at 5:20 | answer | added | Mohammad Alaggan | timeline score: 17 | |
| Nov 4, 2010 at 5:05 | history | edited | gabgoh | CC BY-SA 2.5 |
edited body
|
| Nov 4, 2010 at 4:44 | history | edited | gabgoh | CC BY-SA 2.5 |
added 20 characters in body
|
| Nov 4, 2010 at 4:34 | history | asked | gabgoh | CC BY-SA 2.5 |