From a neural network with 11 input variables, one hidden layer with three hidden units, and one output variable, how many parameters (weights and biases) needs to be estimated by this model?

40

Explanation:

Given that:

A neural network with 11 input variables possess;

one hidden layer with three hidden units; &

one output variable

For every input, a variable must go to every node.

Thus, we can calculate the weights of weight with respect to connections to input and hidden layer by using the formula:

= ( inputs + bias) × numbers of nodes

= (11 + 1 ) × 3

= 12 × 3

= 36 weights

Also, For one hidden layer (with 3 nodes) and one output

The entry result for every hidden node will go directly to the output

These results will have weights associated with them before computed in the output node.

Thus; using the formula

= (numbers of nodes + bais) output, we get;

= ( 3+ 1 ) × 1

= 4 weights

weights with respect to input and hidden layer total = 36

weights with respect to hidden and output layer total = 4

Finally, the sum of both weights is = 36 + 4

= 40