Timeline for What is required for neural network to approximate discontinuous function?
Current License: CC BY-SA 4.0
10 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Dec 31, 2023 at 0:27 | answer | added | wjktrs | timeline score: 0 | |
| Dec 30, 2023 at 18:43 | comment | added | Dikran Marsupial | As well as conditions on the network, there will also be a requirement that the data are sufficiently dense to resolve the discontinuity with the required accuracy. | |
| Dec 30, 2023 at 16:29 | answer | added | Hải Sơn Lê | timeline score: -2 | |
| Feb 18, 2021 at 10:21 | answer | added | Bo Tian | timeline score: 2 | |
| Jan 3, 2021 at 2:25 | comment | added | user76284 | Related: stats.stackexchange.com/questions/311693/… | |
| Sep 2, 2018 at 14:45 | comment | added | Rodolphe | Considering that neural networks are able to approximate any Boolean function (AND, OR, XOR, etc.) It should not be a problem, given a suitable sample and appropriate activation functions, to predict a discontinuous function. Even a pretty simple one-layer-deep network will do the job with arbitrary accuracy (correlated with the number of neurons in hidden layer). I know that for sure because I made almost the same test some years ago (but I do not have the files now, answering from phone) | |
| Sep 1, 2018 at 1:16 | answer | added | Sycorax♦ | timeline score: 10 | |
| Sep 1, 2018 at 0:49 | answer | added | Don Walpola | timeline score: 1 | |
| Aug 31, 2018 at 23:30 | review | First posts | |||
| Sep 1, 2018 at 0:54 | |||||
| Aug 31, 2018 at 23:25 | history | asked | Conv | CC BY-SA 4.0 |