Tensorflow
Below I've embedded a neural network classifier rendered in a Tensorflow playground. There are a variety of knobs, buttons, numbers, and pictures on the interface; I will explain what they all mean as we go along. For now though, let's define our primary goal throughout this tutorial...
Our primary task is to train neural nets to classify items into 1 of 2 categories. Here we represent those categories as either an orange dot or blue dot. You can think of these dots as CASE and CTRL participants in an Alzheimer's Disease (AD) study. For the heck of it, say blue dots represent CASE and orange dots represent CTRL.
In this first example, let's say...
- dim-1 (x-axis): braak score
- dim-2 (y-axis): age
Notice the dots form clusters. If you were asked to draw a line on this plane, to separate these two clusters, it could be easily done. Our brain's neural nets have already solved the the spatial problem. Now let's see if an artificial neural net can solve the same problem.
Go ahead and click the blue start button below; let it run for about 500 epochs (~5 seconds), then click pause.
{{#widget:Tensorflow1}}
How'd it do? Does one neuron, with a single feature input (the value of each dot in the first dimension, the x-axis) perform well in the separation task? If it did well, an orange background should have formed behind the orange dots, and a blue background behind the blue dots.
The color gradient along the surface (under the dots) can be understood as the neural net's prediction 'confidence' at that given coordinate. More directly, it is the value spit-out by the activation function of the 'output layer'. Here, since we only have a single layer, our hidden 'hidden layer' and 'output layer' are one in the same. The output function of our neuron is known as the tanh function.
The tanh function is an extremely common choice for an output function in artificial neural network machine learning frameworks because it yields a nice sigmoid shape, and no matter the magnitude of its inputs, the output from the tanh function is bounded between { 0 : 1}. These are very desirable properties for neural net nodes. Here you see the tanh function evaluated across various x-dim inputs...
Tanh produces a sigmoid output over the range {-2 : 2}, and automatically evaluates to exact values when its argument is the natural logarithm. Speaking of the natural log, that is another very common choice of output function for the same reasons as tanh.
For now, let's not belabor the point that our neuron (and in going forward, all our neurons) are using the tanh function. Maybe just keep this in mind if you're wondering what sorts of numbers are travelling along the axons of these neurons, and ultimately those colored gradients underneath the dots.
Continue to TensorFlow Tutorial Page 2