4.16.1 Top new questions this week

4.16.1.1 Crack CAPTCHA using deep learning

https://mathematica.stackexchange.com/questions/143691/crack-captcha-using-deep-learning

 
Deep convolutional neural networks are very good at computer vision related tasks. Optical Character Recognition (OCR) is one important branch of computer vision. In fact, the convolution neural  ...
 

- asked by xslittlegrass (50 votes), answered by xslittlegrass (52 votes)

4.16.1.2 Changes to Series[...] in version 11.1

https://mathematica.stackexchange.com/questions/143414/changes-to-series-in-version-11-1

 
I have updated to Mathematica 11.1 and I am shocked to see that the Series-function now works differently: 
 
If I enter 
 
Series[x^2+x^3, {x,0,0}] 
 
 
it will return 
 
x^2+O[x]^3 
 
 
which is just weird. In  ...
 

- asked by Matthias Koenig (16 votes), answered by QuantumDot (3 votes)

4.16.1.3 WLNet format incompatible between version 11.0 and 11.1

https://mathematica.stackexchange.com/questions/143699/wlnet-format-incompatible-between-version-11-0-and-11-1

 
It seems that there are some behavior changes of neural network layers from version 11.0 to 11.1 that leads to the incompatibility of WLNet format. 
 
Consider this example: 
 
Network exported in 11.0 
 
...
 

- asked by xslittlegrass (13 votes), answered by Taliesin Beynon (7 votes)

4.16.1.4 DSolve solution depends on ordering of equation

https://mathematica.stackexchange.com/questions/143726/dsolve-solution-depends-on-ordering-of-equation

 
Bug introduced in 8.0 or earlier and persisting through 11.1.0 or later 
 
 
 
I am trying to obtain the general solution, $G(x_1,x_2,x_3,x_4)$ to the very easy system of PDEs: 
$$ 
\partial_{x_3}  ...
 

- asked by Jeremy Upsal (12 votes), answered by Nasser (8 votes)

4.16.1.5 Why the Callout option makes plotting performance poor

https://mathematica.stackexchange.com/questions/143711/why-the-callout-option-makes-plotting-performance-poor

 
Bug introduced in 11.0 and fixed in 11.1 
 
 
 
The Callout option is a new option for plotting, but the performance is very poor, see bellow testing case. 
 
num = 1000; 
list = N@Sin@Range[num]; 
list2 =  ...
 

- asked by Jerry 11votes

4.16.1.6 Neural machine translation model using deep learning

https://mathematica.stackexchange.com/questions/143609/neural-machine-translation-model-using-deep-learning

 
In Mathematica version 11.1, SequenceAttentionLayer is introduced to model attentions in the recurrent neural network. It has been shown that attention mechanism can greatly improve the quality of a  ...
 

- asked by xslittlegrass (11 votes), answered by Sebastian (8 votes)

4.16.1.7 Backslide in NSolve in V11.1?

https://mathematica.stackexchange.com/questions/143824/backslide-in-nsolve-in-v11-1

 
In V11.1, 
 
NSolve[BesselJ[0, x] == 0 && 0 < x < 20, x] 
 
 
returns no solutions: 
 
 
 
But in V11.0 (and earlier), it returns all solutions: 
 
 
 
Is there anyway to get NSolve to solve this  ...
                                                                                  

                                                                                  
 

- asked by Michael E2 (10 votes), answered by Nasser (8 votes)