Table 2.

Neural network algorithms used in nanopore basecaller.

Algorithms	Descriptions
Hidden Markov models [45]	A stochastic model that makes predictions based only on the previous event and a series of observations.
Recurrent Neural Network (RNN) [46]	A model that allows networks with periodic connections to learn complex tasks that require information to be maintained for fixed or indeterminate periods of time.
Long-short-term memory (LSTM) [47]	A type of recurrent neural network that can be used as a component of a larger network. It has specific input, output, and forgetting gates, which can be implemented to retain or discard information passed from the previous state.
Convolutional Neural Network (CNN) [48]	A neural network model for image classification, which extracts input features through a convolution algorithm.
Connectionist Temporal Classification (CTC) [49]	A convolutional neural network for marking neural network output and scoring of sequence data. It does not require pre-segmented training data and post-processed output