| Grp. . | Kernel (K) . | Stride (S) . | Encoder layers . | . | . | . | . | Decoder layers . | . | . | . |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | 11 | 11 | Conv1D(32) | FC | FC(μ,σ) | FC | Conv1D | ||||
| 31 | 31 | Conv1D(32) | FC | FC(μ,σ) | FC | Conv1D | |||||
| 51 | 51 | Conv1D(32) | FC | FC(μ,σ) | FC | Conv1D | |||||
| 2 | 11 | 3 | Conv1D(32) | FC | FC(μ,σ) | FC | Conv1D | ||||
| 11 | 3 | Conv1D(32) | Conv1D(64) | FC | FC(μ,σ) | FC | Conv1D(32) | Conv1D | |||
| 11 | 3 | Conv1D(32) | Conv1D(64) | Conv1D(128) | FC | FC(μ,σ) | FC | Conv1D(64) | Conv1D(32) | Conv1D |
| Grp. . | Kernel (K) . | Stride (S) . | Encoder layers . | . | . | . | . | Decoder layers . | . | . | . |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | 11 | 11 | Conv1D(32) | FC | FC(μ,σ) | FC | Conv1D | ||||
| 31 | 31 | Conv1D(32) | FC | FC(μ,σ) | FC | Conv1D | |||||
| 51 | 51 | Conv1D(32) | FC | FC(μ,σ) | FC | Conv1D | |||||
| 2 | 11 | 3 | Conv1D(32) | FC | FC(μ,σ) | FC | Conv1D | ||||
| 11 | 3 | Conv1D(32) | Conv1D(64) | FC | FC(μ,σ) | FC | Conv1D(32) | Conv1D | |||
| 11 | 3 | Conv1D(32) | Conv1D(64) | Conv1D(128) | FC | FC(μ,σ) | FC | Conv1D(64) | Conv1D(32) | Conv1D |
Two groups of ConvNet-VAEs are employed in this study. In Group 1, the encoders and decoders of ConvNet-VAEs comprise only one 1D convolutional layer, with increasing size of the kernel (K) and stride (S) of filters (from 11 to 51). In the second group, the ConvNet-VAEs contain an increasing number of 1D convolutional layers (from 1 to 3), with fixed kernel size and stride.
| Grp. . | Kernel (K) . | Stride (S) . | Encoder layers . | . | . | . | . | Decoder layers . | . | . | . |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | 11 | 11 | Conv1D(32) | FC | FC(μ,σ) | FC | Conv1D | ||||
| 31 | 31 | Conv1D(32) | FC | FC(μ,σ) | FC | Conv1D | |||||
| 51 | 51 | Conv1D(32) | FC | FC(μ,σ) | FC | Conv1D | |||||
| 2 | 11 | 3 | Conv1D(32) | FC | FC(μ,σ) | FC | Conv1D | ||||
| 11 | 3 | Conv1D(32) | Conv1D(64) | FC | FC(μ,σ) | FC | Conv1D(32) | Conv1D | |||
| 11 | 3 | Conv1D(32) | Conv1D(64) | Conv1D(128) | FC | FC(μ,σ) | FC | Conv1D(64) | Conv1D(32) | Conv1D |
| Grp. . | Kernel (K) . | Stride (S) . | Encoder layers . | . | . | . | . | Decoder layers . | . | . | . |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | 11 | 11 | Conv1D(32) | FC | FC(μ,σ) | FC | Conv1D | ||||
| 31 | 31 | Conv1D(32) | FC | FC(μ,σ) | FC | Conv1D | |||||
| 51 | 51 | Conv1D(32) | FC | FC(μ,σ) | FC | Conv1D | |||||
| 2 | 11 | 3 | Conv1D(32) | FC | FC(μ,σ) | FC | Conv1D | ||||
| 11 | 3 | Conv1D(32) | Conv1D(64) | FC | FC(μ,σ) | FC | Conv1D(32) | Conv1D | |||
| 11 | 3 | Conv1D(32) | Conv1D(64) | Conv1D(128) | FC | FC(μ,σ) | FC | Conv1D(64) | Conv1D(32) | Conv1D |
Two groups of ConvNet-VAEs are employed in this study. In Group 1, the encoders and decoders of ConvNet-VAEs comprise only one 1D convolutional layer, with increasing size of the kernel (K) and stride (S) of filters (from 11 to 51). In the second group, the ConvNet-VAEs contain an increasing number of 1D convolutional layers (from 1 to 3), with fixed kernel size and stride.
This PDF is available to Subscribers Only
View Article Abstract & Purchase OptionsFor full access to this pdf, sign in to an existing account, or purchase an annual subscription.
