NNMLPVisualizer

NNMLPVisualizer

new NNMLPVisualizer(config)

Source:
Parameters:
Name Type Description
config Object
Properties
Name Type Attributes Description
container HTMLElement | string

— Container-Element oder ID
callbacks Object <optional>

— Event-Callbacks

Properties
Name Type Attributes Description
onInputCellClick function <optional>

— (cellIndex)
onHiddenNeuronHover function <optional>
onHiddenNeuronLeave function <optional>

Members

callbacks :Object

Source:
Type:
  • Object

container :HTMLElement

Source:
Type:
  • HTMLElement

lastSnapshot :Object|null

Source:
Type:
  • Object | null

svg :SVGSVGElement|null

Source:
Type:
  • SVGSVGElement | null

topology :Array.<number>

Source:
Type:
  • Array.<number>

viewport :SVGGElement|null

Description:

  • Viewport group for zoom/pan

Source:

Viewport group for zoom/pan

Type:
  • SVGGElement | null

Methods

applyWeightFilter(percentage)

Description:

  • Hides connections whose absolute weight falls below the given percentile. 100% = show all, 0% = hide all.

Source:
Parameters:
Name Type Description
percentage number

— 0–100

destroy()

Description:

  • Destroys the visualizer and cleans up event listeners.

Source:

highlightInputWeights(weights)

Description:

  • Highlights input rects by weight magnitude (receptive field visualization).

Source:
Parameters:
Name Type Description
weights Array.<number>

— 9 weights from a hidden neuron

init(topology)

Description:

  • Initialisiert das SVG für eine gegebene Topologie.

Source:
Parameters:
Name Type Description
topology Array.<number>

— z.B. [9, 36, 9]

resetInputHighlight(grid)

Description:

  • Resets input highlights back to board state.

Source:
Parameters:
Name Type Description
grid Array.<number>

— TTTRegularBoard.grid

updateFromSnapshot(snapshot)

Description:

  • Updates the visualization from a NeuralNetwork snapshot.

Source:
Parameters:
Name Type Description
snapshot NetworkSnapshot

updateInputBoard(grid)

Description:

  • Updates input rects from TTTRegularBoard.grid. grid values: CELL_EMPTY=0, PLAYER1=1, PLAYER2=2

Source:
Parameters:
Name Type Description
grid Array.<number>

— 9 values from TTTRegularBoard.grid

updateOutputHeatmap(predictions, grid)

Description:

  • Updates output rects as a heatmap of NN predictions.

Source:
Parameters:
Name Type Description
predictions Float64Array

— Softmax output (9 values)
grid Array.<number>

— Board grid to dim occupied cells