Emerging Technologies in Computing by Unknown

Author:Unknown
Language: eng
Format: epub
ISBN: 9783030600365
Publisher: Springer International Publishing

4 Comparative Analysis

We have compiled our results based on the accuracy of the model versus the number of features used. The CNN model using MFCC as features gave a very high accuracy for all number of features as illustrated in Fig. 11(a).

Fig. 11.(a) CNN accuracy for Mel spectrogram vs MFCC, (b) LSTM accuracy for Mel spectrogram vs MFCC.

For the LSTM model, Mel Spectrogram features provided average higher accuracy for any number of features as illustrated in Fig. 11(b).

After applying the feature reduction algorithm and training it on the 1D-CNN model, it has been observed that PCA coefficients exhibit very high accuracy among all the other techniques. T-SNE transformation demonstrates the accuracy around 49–50%; whereas the k-PCA shows a moderate accuracy around 72%–75% for the features as illustrated in Fig. 12(a).

Fig. 12.(a) CNN accuracy metrics for PCA vs k-PCA vs t-SNE, (b) LSTM accuracy metrics for PCA vs k-PCA vs t-SNE.

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.