MFOC-CliqueNet: A CliqueNet-Based Optimal Combination of Multidimensional Features Classification Method for Large-Scale Laser Point Clouds

Figure 5

Comparison of different settings: (a) comparison of different initial learning rate. The initial learning rate of 1e − 3 is more conducive to the improvement of classification accuracy. (b) Comparison of different optimizer; optimizer moment is better than optimizer stochastic gradient descent (SGD) and adaptive moment estimation (adam).

(a)

(b)