A Sparse Gating Convolutional Recurrent Network for Traffic Flow Prediction

Figure 8

The loss curve of the model on the TaxiNYC dataset. (a): ConvGRU, SConvGRU, and SConvGRU + on the training set. (b) ConvGRU, SConvGRU, and SConvGRU + on the validation set. (c) ConvLSTM, SConvLSTM, and SConvLSTM + on the training set. (d) ConvLSTM, SConvLSTM, and SConvLSTM + on the validation set.

(a)

(b)

(c)

(d)