Table 6 Night/day class and domain test ROCAUC of logistic regression on \(z^c\) and \(z^d\) using 30 test subjects.
From: Tackling inter-subject variability in smartwatch data using factorization models
Model | Latent space | Class accuracy | Domain accuracy | ||
|---|---|---|---|---|---|
MLP | \(z^c\) | 76.81 | 9.12 | ||
FAE | \(z^c\) | 80.23 | 3.81 | ||
\(z^d\) | 76.12 | 4.53 | |||
GFAE (\(m^d=0\)) | \(z^c\) | 82.21 | 10.24 | ||
\(z^d\) | 76.71 | 11.01 | |||
GFAE (\(m^d\ne 0\)) | \(z^c\) | 83.02 | 12.22 | ||
\(z^d\) | 73.10 | 14.70 | |||
TFAE | \(z^c\) | 84.44 | 17.74 | ||
\(z^d\) | 74.31 | 20.11 | |||
