Fig 1.
CNN architecture having two hidden layers, followed by a dense and output layer. The input vector is given to the 1st convolution (hidden) layer, and output is received via the output layer as distribution of softmax function.
Fig 2.
LSTM architecture having two hidden layers, followed by a dense and output layer. The input vector is given to the 1st convolution (hidden) layer and output is received via the output layer as distribution of softmax function.
Fig 3.
Different classes of GSE6186 gene expression.
X-axis denotes the time interval and Y-axis represents the corresponding gene expression value. (A): Maternal gene expression. (B): Transient gene expression. (C): Activated gene expression.
Table 1.
Accuracy of all methods on different datasets.
All accuracy value is mentioned in percentage (%).
Table 2.
F1 score of all methods on different datasets.
All F1 score value is mentioned in percentage (%).
Table 3.
Accuracy & F1 score of STEM on all dataset.
All value is mentioned in percentage (%).
Table 4.
RMSE value of all methods on different datasets.
RMSE value of different methods for different test percents are grouped together and best RMSE values are highlighted.