Timeline for Advantages of a back-propagation neural network over other function approximation methods
Current License: CC BY-SA 2.5
15 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Dec 28, 2009 at 14:01 | vote | accept | Bruno Reis | ||
| Dec 27, 2009 at 10:50 | answer | added | Alekk | timeline score: 0 | |
| Dec 27, 2009 at 7:59 | answer | added | David Bar Moshe | timeline score: 4 | |
| Dec 27, 2009 at 4:51 | comment | added | Harry Gindi | This question should be closed and asked again with the relevant changes. Please rethink your question before reposting it. (I don't actually have the power to do this, but it seems reasonable.) | |
| Dec 27, 2009 at 4:43 | answer | added | Steve Huntsman | timeline score: 2 | |
| Dec 27, 2009 at 3:14 | comment | added | user577 | I'm not sure what you mean by the input/output pairs won't change. Isn't the whole point of these kinds of algorithms to classify the inputs given some assumptions on how the input varies? So if the input is not varying then this just reduces to a linear algebra problem. | |
| Dec 27, 2009 at 2:14 | comment | added | Qiaochu Yuan | Pattern recognition is a very different question from the question you originally asked. A good handwriting recognition algorithm would require many additional properties of the function f than the properties you asked of it and I am not sure that this problem is written clearly enough that people can help you with what you want. | |
| Dec 27, 2009 at 2:04 | comment | added | Bruno Reis | @Qiaochu Yuan: I've tried to clarify it a bit. | |
| Dec 27, 2009 at 2:02 | history | edited | Bruno Reis | CC BY-SA 2.5 |
added 968 characters in body; added 1 characters in body
|
| Dec 27, 2009 at 1:50 | comment | added | Qiaochu Yuan | I guess what I really want you to clarify is what you want f for. "Advantage" has no meaning except with respect to some application. | |
| Dec 27, 2009 at 1:29 | comment | added | Bruno Reis | @Qiaochu Yuan: I've edited the question. I want a mapping that is at least continuous around the input/output pairs. | |
| Dec 27, 2009 at 1:28 | history | edited | Bruno Reis | CC BY-SA 2.5 |
added 78 characters in body
|
| Dec 27, 2009 at 1:19 | comment | added | Qiaochu Yuan | What is a "mapping" for you? Linear? Differentiable? Continuous? | |
| Dec 27, 2009 at 1:18 | comment | added | Gjergji Zaimi | What other methods of approximation did you have in mind? | |
| Dec 27, 2009 at 1:10 | history | asked | Bruno Reis | CC BY-SA 2.5 |