W-Net structure. A 64 × 64 × 64 sample of median-subtracted TESS data built using the process described in Section 2 is the input to the neural network. The data undergo our custom adaptive normalization process, described in Section 4, where the model learns the parameters of a mixture of logistic cumulative distribution functions (CDFs) that provide a learned transformation of the input data to the [0,1] range. Convolutional blocks (CBs) are given by their output dimensions and described further in Figure 5. Activations internal to the convolutional blocks use the exponential linear unit (ELU) function (D.-A. Clevert et al. 2015), while the final and innermost 64 × 64 × 64 × 1 convolutions are activated with the sigmoid function.