Skip to main content
AI Perspectives & Advances

Table 4 Training setting for experiment 1

From: CLeaR: An adaptive continual learning framework for regression tasks

Hyperparameter

Value

Optimizer

Adam

Epochs in phase 1

512

Epochs in phase 2

512

Batch size in phase 1

32

Batch size in phase 2

16

Online EWC γ

0.9

Online EWC Îğ

200

  1. Epochs denote the number of iterations for training the autoencoder. The batch size is the size of a mini-batch. Phase 1 and phase 2 indicate the warm-up phase and the update phase respectively. The experiments of training baseline model adopt the setting of the phase 1
Back to article page