Skip to content

Support multi-digit classification (more than 0 and 1) #3

Description

@provandal

Currently the network architecture is fixed at 81 → 24 → 2, supporting only binary digit recognition (0 vs 1).

Proposed enhancement:

  • Allow configurable output neuron count (e.g., 10 for digits 0-9)
  • Update the label selector UI to support multiple classes
  • Adjust the training data format to support multi-class one-hot encoding
  • Consider a larger hidden layer or configurable architecture

This would make the tool useful for exploring how network capacity relates to task complexity.

Metadata

Metadata

Assignees

No one assigned

    Labels

    enhancementNew feature or request

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions