This prototype illustrate the network’s layers, neuron connections, and data flow, offering a tangible cross-sectional representation of the MLP. The physical prototype serves as an essential early-stage tool for conceptualizing the MLP network structure before advancing to more detailed reality sketches and sequence designs. Its primary purpose was to illustrate the network’s layers, neuron connections, and data flow, offering a tangible cross-sectional representation of the MLP.
The reality sketches accurately capture what the user experiences in a feedforward MLP. This visual representation demonstrates a step-by-step exploration of the network’s computations. In the first sketch, the user (as a MLP neuron) is focused on viewing an incoming vector in the hidden layer, observing their values before weight update. Each value is visualized within the user’s reach.