Relevant research (2018-2022)
■ [4]: input layer, multiple CNN, BLSTM, attention layer
– Accuracy 84,2%
– With DT, MLP, RF, accuracy 72 − 78%
■ [5]: AE (115>50 features), sparse AE (50>10), LR (10>2)
– Binary classification only
– Accuracy 87,2%
■ [6]: Best of performance of all with LSTM, DCNN, Denoising and Contractive AE
– Accuracy 89% 𝐿𝑆𝑇𝑀 , 81 − 85% 𝐴𝐸
– RF, DR, k-NN, MLP algorithms → accuracy 74 − 82%
30