Ciena Corporation

Yan Liu

David Côté

Thomas Triplet

Yang Yu

Qunbi Zhuge

Emil Janulewicz

Arslan Shahid

Merlin Davies

Kevin Kim

Systems and methods of automated feature selection and pattern discovery of multi-variate time-series include obtaining a multi-variate times-series from a network; preprocessing the multi-variate times-series to account for sampling intervals and missing data in the multi-variate times-series; determining a distance matrix for the multi-variate times-series which estimates correlation among features in the multi-variate times-series; performing clustering on the distance matrix; reducing dimensionality of the multi-variate times-series based on the clustering to provide a lower-dimensionality time-series; and providing the lower-dimensionality time-series to one or more applications configured to analyze the multi-variate times-series from the network, wherein the lower-dimensionality time-series provides similar information as the multi-variate time-series with fewer dimensions thereby improving computational complexity of the one or more applications.

Systems and methods for automated feature selection and pattern discovery of multi-variate time-series

Technology for link parameter identification in an optical communications network is described. A first trained artificial neural network (ANN) may be applied to first input values representative of nonlinear noise in a signal received at a receiver from a transmitter over a link in the optical communications system, thereby generating first output values. A second trained ANN may be applied to second input values comprising the first output values and one or more known parameters of the link, thereby generating second output values. One or more link parameter estimates may be identified based on the second output values. In some examples, the first trained ANN has an architecture specialized for two-dimensional image recognition and therefore suitable for the image-like properties of the first input values. For example, the first trained ANN may comprise a deep residual learning network (ResNet) or a Convolution Neural Network (CNN).

Machine learning for link parameter identification in an optical communications system

Growing community of inventors

Montreal, Canada

Yan Liu