Active Dendrites vs Sparse Representations ablation study

[karangrewal]

Configurations for continual learning ablation studies on permutedMNIST which include networks with

• active dendrites but no sparse representations (ReLU instead of k-Winners) and
• sparse representations but no active dendrites.