VPIphotonics is showing its new machine learning (ML) framework to design and implement deep neural networks for optical communication systems and photonic devices at ECOC 2022.
The VPItoolkit™ ML Framework is an add-on to any of the simulation tools of VPIphotonics Design Suite™, enabling the implementation and design of deep neural networks (DNN) for various applications, such as equalisation and nonlinearity mitigation for optical systems, device characterisation, evaluation and inverse design of photonic devices.
The aim of VPItoolkit ML Framework is to build a model that makes predictions based on evidence in the presence of uncertainty, by collecting known training data sets, which are used to train the supervised DNN model. The framework provides an open-source Python-based DNN with an intuitive interface to manipulate model parameters and convergence constraints. Additionally, it allows for the deployment of custom-made ML algorithms by the user.
Seamless manipulation of digital, electrical and optical signals is feasible, using the framework’s flexible Data Extractors and Model Loaders. With user-friendly access to the DNN hyperparameters, fast optimisation can be performed for model performance enhancement.
VPItoolkit ML Framework supports the storage of large, complex, heterogeneous data in the open-source file format Hierarchical Data Format version 5 (HDF5).
For more information, visit www.vpiphotonics.com