PureBasic Forums - English

In answer to my own question, yes the model is overfitting. When that happens one of the things one should try adjusting is the Learning Rate - start coarse and go granular (whereas I started way too granular). Check the error on the validation data on each iteration and look for the lowest level ...

This is a little 2 layer neural net from scratch in 'simple code' which is supposed to produce the result y=(a*b)+c when presented with three inputs a,b,c where 0<(y,a,b,c)<1. It consists of 3 neurons in a hidden layer and one neuron at the output. Each of the neurons has a sigmoid activation ...

Well one of the by-products of this activity has been the listing of a new sequence A316742 at https://oeis.org/.
Once again thanks for the suggestion to raise this at http://www.mersenneforum.org/index.php...one never knows where a line of investigation can lead

Thanks so much applePi, Comtois and DK_Peter for your invaluable assistance with 3D polygon programming (in which I had no experience prior)!
It helped me turn into reality the idea of generating a changing, in-sync sound waveform from the motion of the vertices of a 3D object :D
The app code and ...

I publish here the full code for my sound synthesis app 'Rotatosound' (and following that the Portaudio code which is an 'Include' in the app). It was done with PB version 5.31.
It generates a changing sound waveform from the motion of the vertices of rotating 3D objects.
It runs fine on my Win XP ...

Just resolved:
How to swap the mesh of one object with another 'on the fly'?
By simply calling a Procedure like the following upon change of a GadgetState. Was getting a blank in the image window previously as had not attached the mainnode to the newly created mesh :oops:
Procedure Tetrahedron ...

Thanks applePi. I've had a go at creating meshes with Paul Bourke's published co-ordinates for a variety of polygons, and they've worked out well. In case anyone else finds them handy:
http://paulbourke.net/geometry/platonic/
http://paulbourke.net/geometry/spacefill/
Another question: How to ...

Thanks Comtois, yes I found that pyramid example too not long before you posted! :mrgreen:

In an attempt to try to get the pyramid to look like DK_Peter's transparent cube with the green edges, I commented out a couple of the 'material' lines in the pyramid example and pasted DK_Peter's 'material ...

Thanks for your help! I've modified DK_PETER's code to display the (x,y) pixel coords of the 8 cube vertices (or at least I think I have). Applied CameraProjection to the 3D coords of the nodes like this:
Debug Str(x) + ";" + StrF(CameraProjectionX(0, NodeX(n(x)), NodeY(n(x)), NodeZ(n(x ...

Thanks for the tip jack, I am indeed getting good engagement and input from members of mersenneforum re this topic:
http://www.mersenneforum.org/showthread.php?t=23501

Wondering if someone would be so kind as to suggest mods to this rotating cube code for the following:
-make the cube 'transparent' (such that all vertices are visible at all times)
-enable rotation in 'different directions' from the same viewing position
-capture the (x,y) pixel co-ordinates of the ...

With the sequence grapher, by changing the initial values from aii.q = 0, bii.q = 1 to aii.q=0, bii.q=0 and observing the resulting sequences, we can summarize the relations as follows:

If T(n) = T(n-1) + 2T(n-2) -2T(n-3) and W(n) = T(n) mod m, a series of related sequences w(n), w'(n), w''(n) are ...

Here's the code for the sequence grapher, with which I came across this interesting phenomenon (substituting NicTheQuick's suggestion for the sequence generation rather than using my original cumbersome string-based routines - thanks very much for that!)
It features sliders for changing the values ...

I haven't come across this in https://oeis.org or other online search, so thought I'd post it here for a peer review (and share the simple but ineifficient 'big number' code I used in the calculation):
In the 'classical' Tribonacci sequence, each value is the sum of the previous three, like so:
T(n ...

Thanks Wilbert for the effort - your ASM optimized code's been running fine and producing the right results