Key points are not available for this paper at this time.
One of the most important tasks in a general computational framework for biomedical data analysis is the categorization of DNA sequences. Several machine learning techniques have been used in recent years to successfully complete this job. In any case, the feature selection procedure continues to be the primary source of difficulties for the issue. Sequences lack explicit characteristics, and the primary issue of high dimensionality is introduced by the generally employed representations. It is true that machine learning techniques for supervised classification problems heavily rely on the feature extraction stage, and that important aspects of the objects to be classified must be identified and measured in order to construct a suitable representation. Neural deep learning architectures, often known as deep learning models, have demonstrated the ability to automatically extract meaningful features from input patterns. Two distinct encoding techniques—k-mer and label encoding, are presented as input to the suggested model of CNN classification along with embedded layer to represent DNA sequences. When compared to various encoding techniques, the suggested CNN hybrid model with label encoding had the better classification accuracy, at 97.22%
Nerkar et al. (Sat,) studied this question.