Skip to main content

Table 2 Selected optimal hyper-parameters as number (No.) of hidden neurons, optimal (Opt.) window size, average test accuracy values, and the corresponding standard deviations for each model

From: Temporal social network reconstruction using wireless proximity sensors: model selection and consequences

  unreconstr. naïve HMM BiLSTM-bin BiLSTM-logi BiLSTM-RSSI
No. hidden neurons 64 16 32
Opt. window size 0 6 6 27 25 27
Average accuracy 77.28 % 83.36% 84.25% 88.34% 89.02% 90.03%
Standard deviation 14.91% 12.11% 11.67% 9.55% 9.36% 8.99%