Timeline for How to use CNN for making predictions on non-image data?
Current License: CC BY-SA 4.0
13 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Mar 27, 2023 at 16:26 | comment | added | userrandrand | The company website itself says the following: "... direct modifications of the Transformer architecture. Of course, the neural networks of DeepL also contain parts of this architecture, such as attention mechanisms. However, there are also significant differences in the topology of the networks". So a plausible guess is that they started with CNN's at some point and then made some architectural designs that integrate the attention layers. | |
| Mar 27, 2023 at 16:21 | comment | added | userrandrand | Some websites claim that DeepL uses a CNN but I did not find this information on the company's website itself. Here is one website that makes this claim: blog.meinrad.cc/en/…. This website: slator.com/… also mentions that DeepL seemed to use that in the past but later mentions someone that said “I’d be surprised if DeepL is still using CNNs,” | |
| S Apr 13, 2021 at 15:54 | history | suggested | Faizy | CC BY-SA 4.0 |
fixed the grammar
|
| Apr 13, 2021 at 12:42 | review | Suggested edits | |||
| S Apr 13, 2021 at 15:54 | |||||
| Jun 23, 2019 at 20:47 | vote | accept | JChat | ||
| Feb 18, 2019 at 23:02 | comment | added | tsbertalan | @DuttA Yeah, I guess if filter size=image size, and you use "valid" padding, the output will be 1x1xM, after which further convolutions will just be 1x1xN, then 1x1xO, etc as I said, where M, N, O, etc. are equivalent to the widths of successive layers in a normal MLP. I mentioned 1x1 convolutions since this is a common way that MLP-like results are achieved with CNNs pointwise (e.g. in fully-convolutional NNs). | |
| Feb 12, 2019 at 5:54 | comment | added | user9947 | @tsbertalan CNN literature specify filter size, stride, etc. If filter size=image size then it is a NN (by NN I mean a fully connected NN). Otherwise speaking rnn, CNN are all NN's. Here I meant a FCNN by NN. | |
| Feb 12, 2019 at 4:20 | comment | added | tsbertalan | @DuttaA I think "CNNs are just a special case of NNs" would be more correct. Unless you're thinking of 1x1 convolutions of a 1x1 input "image" of 1972 "channels".. | |
| Feb 8, 2019 at 22:08 | history | edited | malioboro | CC BY-SA 4.0 |
added 99 characters in body
|
| Feb 8, 2019 at 22:05 | comment | added | malioboro | hmm that's true, that's why I said "it's recommended to use CNN only on data that have spatial features", I think I'll add your comment to give more emphasis | |
| Feb 8, 2019 at 10:49 | comment | added | user9947 | It still might work without spatial features considering CNN s NNs are just a special case of CNNs. | |
| Feb 8, 2019 at 10:31 | vote | accept | JChat | ||
| Feb 8, 2019 at 17:41 | |||||
| Feb 8, 2019 at 0:52 | history | answered | malioboro | CC BY-SA 4.0 |