{"id":419,"date":"2019-10-14T17:42:11","date_gmt":"2019-10-14T17:42:11","guid":{"rendered":"http:\/\/www.kobakant.at\/false-lies\/?p=419"},"modified":"2019-11-04T12:02:44","modified_gmt":"2019-11-04T12:02:44","slug":"e-textile-ann-forward-prop","status":"publish","type":"post","link":"https:\/\/www.kobakant.at\/false-lies\/e-textile-ann-forward-prop\/","title":{"rendered":"e-textile ANN forward prop"},"content":{"rendered":"\n

Here is the first prototype of e-textile ANN. It is only 3 neurons, hard-coded weights and biases at the moment. The input layer is simply 2 potentiometers (2 inputs), connected to 2 neurons on a hidden layer, 2 inputs each, and 1 neuron for output layer. It is basically the same as the ATTINY simulation on the breadbaord, but made onto textile as embroidered circuit.<\/p>\n\n\n\n\"textile<\/a>