“…We show the applicability of the TL16-8/4 multiplier and the TL8-4/3 multiplier in the image smoothing and the image classification with convolutional neural networks. In both applications, we compare the TL16-8/4 multiplier with the state-of-the-art approximate multipliers Mitchell-trunc6-C1, Mitchell-trunc8-C1, Mitchell-trunc6-C1UB, Mitchell-trunc8-C1UB [12], ILM-AA [15], ALM-SOA11 [23], DR-ALM4 [8], and RAD1024 [9] and the TL8-4/3 multiplier with the Mitchell-trunc5-C1, Mitchell-trunc5-C1UB [12], ALM-SOA4, ALM-SOA6 [23], DR-ALM3, DR-ALM4 [8] and Ax8_3 [32] multipliers.…”