Artificial Neural Networks
Artificial neural networks (ANNs) are computational predictive
analytical systems inspired by the human nervous system. These consist
of a high number of computational nodes (called neurons) that spread
across various layers. In simplest terms, it is comprised of three basic
layers with distinct functions.
- An input layer, in which data (usually multidimensional vector) is
ingested and distributed to the hidden layer.
- A hidden layer, which makes decisions to assign random weights within
each node to determine if it detriments or improves the output
(referred to as learning). When multiple hidden layers are stacked
together to perform a complex pattern recognition task, it is referred
to as DL. There are two fundamental learning approaches, supervised
and unsupervised learning. Supervised learning is most often used for
image-based pattern recognition tasks. In this form, for every
training set (e.g., a single digital ECG), input vectors are
associated to one or more pre-assigned labels 13. In
contrast unsupervised learning does not have pre-selected labels, but
rather seeks to find clusters of salient features from the data
itself.
- An output layer where, in the case of supervised learning, a system
output is compared to the preset label output and backpropagated to
the previous layers with the goal of reducing classification error as
much as possible by tuning the weights.
Convolutional neural networks (CNN) are similar to traditional ANN, but
are particularly well suited for image recognition tasks, and thus
reduce the parameters required to set up the model. The capability of
these models to analyze subtle details from abstract data is remarkable
and far superior to humans. More details on these models are described
elsewhere.14