A miRNA Target Prediction Model Based on Distributed Representation Learning and Deep Learning

Figure 5

Different embedding influences the deep-learning model on accuracy. (a) Shown is the use of one-hot and word2vec (2, 4, 10, 20, 30, 50, and 100-dimensions) as the input of the CNN1d model. (b) The same test using the BiLSTM model.

(a)

(b)