Function updateDecisionBoundary

src/playground.ts:796–838

* Given a neural network, it asks the network for the output (prediction) * of every node in the network using inputs sampled on a square grid. * It returns a map where each key is the node ID and the value is a square * matrix of the outputs of the network for each input in the grid respectively

(network: nn.Node[][], firstTime: boolean)

 * matrix of the outputs of the network for each input in the grid respectively.
 */
function updateDecisionBoundary(network: nn.Node[][], firstTime: boolean) {
  if (firstTime) {
    boundary = {};
    nn.forEachNode(network, true, node => {
      boundary[node.id] = new Array(DENSITY);
    });
    // Go through all predefined inputs.
    for (let nodeId in INPUTS) {
      boundary[nodeId] = new Array(DENSITY);
    }
  }
  let xScale = d3.scale.linear().domain([0, DENSITY - 1]).range(xDomain);
  let yScale = d3.scale.linear().domain([DENSITY - 1, 0]).range(xDomain);

  let i = 0, j = 0;
  for (i = 0; i < DENSITY; i++) {
    if (firstTime) {
      nn.forEachNode(network, true, node => {
        boundary[node.id][i] = new Array(DENSITY);
      });
      // Go through all predefined inputs.
      for (let nodeId in INPUTS) {
        boundary[nodeId][i] = new Array(DENSITY);
      }
    }
    for (j = 0; j < DENSITY; j++) {
      // 1 for points inside the circle, and 0 for points outside the circle.
      let x = xScale(i);
      let y = yScale(j);
      let input = constructInput(x, y);
      nn.forwardProp(network, input);
      nn.forEachNode(network, true, node => {
        boundary[node.id][i][j] = node.output;
      });
      if (firstTime) {
        // Go through all predefined inputs.
        for (let nodeId in INPUTS) {
          boundary[nodeId][i][j] = INPUTS[nodeId].f(x, y);
        }
      }
    }
  }
}

function getLoss(network: nn.Node[][], dataPoints: Example2D[]): number {
  let loss = 0;