Google launches TFQ a machine learning framework for training quantum models

On March 9, local time, Google, along with the University of Waterloo, Volkswagen and others, jointly released TensorFlow Quantum (TFQ), an open source library that can quickly prototype quantum machine learning models. TFQ provides the tools necessary to combine quantum computing and machine learning techniques to control and model natural or manual quantum computing systems.

Google launches TFQ a machine learning framework for training quantum models

The framework enables the construction of quantum data sets, hybrid quantum models and prototypes of classic machine learning models, support for quantum circuit simulators, and training and generation of quantum models.

With the development of quantum computing technology in recent years, the development of quantum machine learning models may make breakthroughs in medicine, materials, sensing and communication, and even have far-reaching effects. So far, however, the industry lacks the research tools to discover quantum machine learning models. The model can process quantum data and execute it on available quantum computers.

In fact, back in October 2017, Google announced the source code for OpenFermion, an open-source quantum computing software that allows users to use its adapted algorithms and equations to run on quantum computers. In October 2019, Google CEO Sundar Pichai announced that the company had achieved quantum supremacy, achieving its first quantum advantage with a newly designed solution.

The launch of TensorFlow Quantum is another step forward following the launch of Microsoft’s Azure Quantum and the milestone success of companies such as Honeywell.

Google launches TFQ a machine learning framework for training quantum models

According to the blog, quantum models can be created through the standard Keras library and the provision of quantum circuit simulators and quantum computing primitives compatible with the existing TensorFlow API.

The framework based on the Python language is described in a paper submitted to the line database platform arXiv on March 6.

“We hope that the framework will provide the research community in quantum computing and machine learning with the necessary tools to explore models of natural and artificial quantum systems and eventually discover new quantum algorithms that may yield quantum advantages,” the paper notes. “In the future, we want to extend the range of custom simulation hardware that can be supported, including THE integration of GPUs and TPU. “

In this paper, the paper details the TensorFlow Quantum software stack, which consists of the open source quantum circuit library Cirq and the machine learning platform TensorFlow.

The paper has more than 20 authors from Google X Labs, the Institute of Quantum Computing at the University of Waterloo, NASA Quantum AI Labs, Volkswagen, and Google Research.

TensorFlow Quantum is understood to have launched the same week as the TensorFlow Dev Summit, the annual meeting of machine learning practitioners. However, Google canceled the offline event because of the continuing effects of the new crown pneumonia.