One of the challenges of implementing a Quantum Machine Learning algorithm (QML) in a quantum computer is initially loading the classical data into the qubits. Although a certain number of qubits can theoretically hold more information using superposition and entanglement than the same number of bits in a classical computer, loading the data so that one can start processing a QLM algorithm has been a big challenge. Although some researchers are looking at Quantum Random Access Memory (QRAM) for this, QC Ware has developed two algorithms they call the Forge Parallel Data Loader and the Forge Optimized Data Loader that they say are a breakthrough and can implement this step very efficiently without using QRAM. For comparison, QC Ware has supplied the table below:
The addition to the QML data loaders, the new version of Forge also includes many other additions including:
- Backend interfaces to hardware on Amazon Braket including the IonQ and Rigetti gate level machines, D-Wave’s Quantum Annealer with improved performance for large problems, NVIDIA GPU’s on Amazon Web Services, and classical computer simulators
- Automatic importing and translation of IBM Qiskit and Google Cirq circuits
- A new library that experts can use to compose circuits
- Optimized Distance Estimation algorithms for QML that will allow for powerful quantum classification and clustering applications
QC Ware will be holding an online webinar on July 22 to discuss the new features. For more information on the new features and the webinar you can view QC Ware’s news release which you can find here.
July 21, 2020