Table 8 Multi-SPP-VAE architecture configurations and optimal hyperparameters under different channel numbers.
From: Quality prediction using multiscale convolutional VAEs for thin plate parts
Dataset | Channels | Conv1d | Multi-SPP | Conv1d | \(\:lr\) | \(\lambda_{kl}\) | \(\lambda_1\) | \(\lambda_2\) |
|---|---|---|---|---|---|---|---|---|
(k=7) | (k=7) | (k=3) | ||||||
A | 27 | In channel =3, | 3 | In channel =9, | 0.0085 | 1.08E-02 | 1.08E-02 | 1.08E-02 |
out channel =3 | out channel =9 | |||||||
54 | In channel =3, | 6 | In channel =18, | 6.10E-03 | 1.00E-02 | 1.00E-02 | 8.13E-02 | |
out channel =6 | out channel =18 | |||||||
81 | In channel =3, | 6 | In channel =18, | 0.0459 | 5.45E-02 | 5.45E-02 | 5.45E-02 | |
out channel =6 | out channel =27 | |||||||
108 | In channel =3, | 6 | In channel =18, | 0.0256 | 3.09E-02 | 3.09E-02 | 0.0601 | |
out channel =6 | out channel =36 | |||||||
B | 27 | In channel =3, | 3 | In channel =9, | 0.0189 | 3.18E-02 | 3.14E-02 | 3.14E-02 |
out channel =3 | out channel =9 | |||||||
54 | In channel =3, | 6 | In channel =18, | 8.60E-03 | 1.14E-02 | 1.14E-02 | 2.98E-02 | |
out channel =6 | out channel =18 | |||||||
81 | In channel =3, | 6 | In channel =18, | 0.0345 | 3.98E-02 | 3.98E-02 | 3.98E-02 | |
out channel =6 | out channel =27 | |||||||
108 | In channel =3, | 6 | In channel =18, | 0.0587 | 5.25E-02 | 5.25E-02 | 5.25E-02 | |
out channel =6 | out channel =36 | |||||||
C | 27 | In channel =3, | 3 | In channel =9, | 0.0071 | 1.00E-02 | 1.00E-02 | 1.00E-02 |
out channel =3 | out channel =9 | |||||||
54 | In channel =3, | 6 | In channel =18, | 6.70E-03 | 1.00E-02 | 1.00E-02 | 1.00E-02 | |
out channel =6 | out channel =18 | |||||||
81 | In channel =3, | 6 | In channel =18, | 0.0081 | 1.00E-02 | 1.00E-02 | 0.3624 | |
out channel =6 | out channel =27 | |||||||
108 | In channel =3, | 6 | In channel =18, | 1.35E-02 | 1.00E-02 | 1.00E-02 | 1.00E-02 | |
out channel =6 | out channel =36 |
