Discrete Cosine Transformation and Temporal Adjacent Convolutional Neural Network-Based Remaining Useful Life Estimation of Bearings

Table 5

Time consuming in model training (unit: s).

Input size/H × W × C	TFR-TACNN	DCT-TACNN
500 × 500 × 2	Overflow	598 (pruned 26.5%)
250 × 250 × 2	1021 (pruned 5.9%)	266 (pruned 20.9%)
100 × 100 × 2	452 (pruned 10.0%)	72 (pruned 17.5%)
50 × 50 × 2	231 (pruned 9.5%)	40 (pruned 24.0%)