Machine Learning {Neural Networking} — Week 2

Your limitation — it’s only your imagination

Welcome to the boring world :-) Today’s topic is theoretical lecture.

Every unique input data into unique neural input layer gives a desired sum of neural output for an input. When you provide an input to each layer the sum and dot product of the neural network output. Remember: Each input to a layer is a forward bias to another neural layer.

Forgive me, I tried to draw it in paint

1. Each circle in the above diagram is called as a neurons.
2. Labels named as Input, Hidden Layer 1, Hidden Layer 2 and Output, you can call it as neuron layer and each layer is called. Each layer has neurons associated with it. The more number of neurons in a layer gives more accurate computation result.

For example: Input layer has two layer, Hidden layers has 4 neuron and finally Output has two neuron.

You can count number of neurons as bias. We totally have 12 neuron or you can call it as 12 biases. Weights are connectivity from each neuron to other neuron in the layer, we have 32 weights.

I do practicals on https://colab.research.google.com , it avoids newbies to avoid unnecessary installation and makes it very flexible to do labs.

Now, let’s talk about multilayer programmable approach.

EACH NEURON IS AN INPUT TO OTHER NEURONS

Code snippet:

input_neurons = [1.2, 5.1, 10.9]

unique_weights_for_each_input_neurons = [9.2, 13.8, 55.7]

unique_bias_for_each_input_neurons = 3

output = input_neurons[0] * unique_weights_for_each_input_neurons[0] + input_neurons[1] * unique_weights_for_each_input_neurons[1] + input_neurons[2] * unique_weights_for_each_input_neurons[2] + unique_bias_for_each_input_neurons

print(output)

________

See you guys next week… Continued.