Today, Google has announced the release of Tensorflow Quantum – for hybrid quantum-classical machine learning. Researchers from Google Research in collaboration with researchers from several Universities have developed this software framework for quantum machine learning.
The goal of the project behind Tensorflow Quantum was to integrate quantum computing algorithms with high-performance quantum circuit simulators and provide primitives that are compatible with existing Tensorflow APIs. As a machine learning framework, Tensorflow Quantum offers high-level abstractions for designing and training both generative and discriminative quantum models that can be deployed on high-performance quantum circuit simulators. It also offers a number of tools to explore these models and it offers researchers the possibility to discover new quantum algorithms.
The framework includes two datatype primitives which are actually needed to integrate TensorFlow with quantum computing hardware: Quantum Circuits and Pauli sum. It also includes utilities for converting quantum circuits to TF tensors and running hybrid quantum-classical optimizations.
Together with the release of TensorFlow Quantum, researchers released a set of notebook tutorials to help users start using the framework. The tutorials can be found here. More in detail about Tensorflow Quantum can be read in the release blog post or in the paper.
