Table 3 The effect of the SSL approach.

From: Self-supervised learning of wrist-worn daily living accelerometer data improves the automated detection of gait in older adults

Model's head

MTL

SimCLR

FC (without non-linearity)

84.67 (0.44)

84.52 (0.38)

FC (with non-linearity)

84.74 (0.51)

84.57 (0.76)

U-Net

83.02 (0.86)

84.28 (0.45)

  1. Performance is reported as the F1 score (standard deviation between different seeds of validation, see the “Methods” section for more details).
  2. FC fully-connected.
Back to article page