Layer Object

Import

const Layer = require('dannjs').layer;

Contructor( type , inputSize , sampleSize , stride )

When you create a pooling layer you need to specify the number of input size, the sample size & the stride.

const l1 = new Layer('avgpool',16,2,2);

Object Properties

• type
  A string representing the type of this layer.

• subtype
  A string representing the sub type of this layer.

• sampleSize
  The size of the 2d sample iterating trough the array.

• stride
  The number of jumps the sample is going to perform for each iteration.

• pickFunc
  The function this pooling layer uses to process a sample as an array into a numeric output. ex: avgpool, maxpool. See available pooling functions Here.

Functions

feed( input , options );

This function feeds data through the layer to obtain an output.

• input
  Takes an array of inputs to feed through the layer.

• options (optional)
  An object including specific properties.

• returns
  Returns an array of outputs

Examples

Here is how you would create a pooling layer & compute some 16 length array of data.

const Layer = require('dannjs').layer;  //nodejs only

const l1 = new Layer('avgpool' , 16, 2, 2);

const input = [
    4, 3, 1, 5,
    1, 3, 4, 8,
    4, 5, 4, 3,
    6, 5, 9, 4
];

const output = l1.feed(input);
console.log(output);

// Output:
// [2.75, 4.5, 5, 5]

You can also use this layer for a non-squarable array by specifing width & height when constructing the layer.

const Layer = require('dannjs').layer;  //nodejs only

const l1 = new Layer('avgpool' , 12, 2, 1, 4, 3);

const input = [
    4, 3, 1, 5,
    1, 3, 4, 8,
    4, 5, 4, 3
];

const output = l1.feed(input);
console.log(output);

// Output:
// [2.75, 2.75, 4.5, 2.75, 4, 4.75]