Skip to main content
. 2023 Jun 6;4(2):48. doi: 10.1007/s43069-023-00224-5

Table 6.

Computational results for comparing LSTM-Opt with different exact methods*

Train Test pred inf time timeimp optgap(%)
c f T T (%) CPX Dpineq LSineq LSTM-Opt Dpineq LSineq LSTM-Opt CPX Dpineq LSineq LSTM-Opt
3 10000 90 360 50 0.0 3602 3609 3602 11.2 1 1 322 0.03 -0.1 -0.1 0.8
5 10000 90 360 50 0.0 3602 3421 3600 20.3 1 1 177 0.05 0.0 0.0 0.9
8 10000 90 360 50 0.0 7.5 720 146 2.5 0 0 3 0.00 0.0 0.0 1.6

*Experiments only include ten test instances and limited with 1-hour time limit due to long solution times