An unsupervised online monitoring method for tool wear ...
An auto-encoder is an unsupervised learning model, which is trained to compute the representation of input to reconstruct inputs as accurately as possible [31, 32].An auto-encoder (Fig. 1) consists of three layers: an input layer, a hidden layer, and an output layer.The hidden layer representation of input vector x is encoded as h(x) by an encoding function f(·), which is …