-
petervr committed patchset 43 of module website to the Lightweight Neural Network CVS repository, changing 1 files.
2009-10-26 07:45:25 UTC by petervr
-
I's been ages since I looked at lwneuralnet, but this feature will be in the next version (whenever that will be ready). It's already in CVS.
2009-10-25 19:08:28 UTC by petervr
-
petervr committed patchset 133 of module lwneuralnet to the Lightweight Neural Network CVS repository, changing 4 files.
2009-10-25 19:04:23 UTC by petervr
-
tEMswm nwfnfvefjwjq, [url=http://cuulnqbtiefb.com/]cuulnqbtiefb[/url], [link=http://exwebzhysfno.com/]exwebzhysfno[/link], http://tlainijsijbz.com/.
2009-09-24 11:56:49 UTC by nobody
-
r1x8im icmnnpyfikyj, [url=http://igwcjugbcaun.com/]igwcjugbcaun[/url], [link=http://zbvldkrdbhdn.com/]zbvldkrdbhdn[/link], http://zkgjbnbmgfzf.com/.
2009-03-12 10:52:20 UTC by nobody
-
This Tracker item was closed automatically by the system. It was
previously set to a Pending status, and the original submitter
did not respond within 14 days (the time period specified by
the administrator of this Tracker).
2008-11-27 02:20:10 UTC by sf-robot
-
petervr committed patchset 132 of module lwneuralnet to the Lightweight Neural Network CVS repository, changing 4 files.
2008-11-17 12:07:01 UTC by petervr
-
Provide a function pointer in Layer structure to an activation function. This allows users of the libary not only to use the sigmoid activation function, but simply define one themselves. I, for example, used a lineair layer as a buffer, passing a bias to the next layer.
typedef struct {
int no_of_neurons;
neuron_t *neuron;
float (*activation_function)(float);
} layer_t;.
2008-11-13 08:30:34 UTC by nobody
-
It's ages ago that I wrote this, but the code (and comments) seem to be consistent. Are you sure this is a bug?
The activation value of node u in layer l is given by sigma(a0 n0 + ... a(k-1) n(k-1) + b) where n0, ..., n(k-1) are the activation values of the nodes in the previous layer (l-1); a0, ..., a(k-1) are the weights; b is the bias.
Now the weights 'belong' to both node u in layer l...
2008-11-12 20:59:12 UTC by petervr
-
When applying a bias and calling net_set_bias the code sets the weight on the previous layer. But connection weights are stored in the next layer, so net_set_weight should store its weights in te next layer to be consistent.
I should change the code in net_set_weight:
net->layer[l].neuron[nu].weight[nl] = weight;
to
net->layer[l+1].neuron[nu].weight[nl] = weight;.
2008-11-11 19:34:50 UTC by nobody
