In short
A VQC is a small quantum circuit with adjustable “knobs”. Each input feature is loaded onto one qubit, the circuit is measured to give a guess, and an ordinary computer turns the knobs until the guesses are right.
A loop between a quantum and a classical computer
Variational Quantum Algorithms (VQAs) are a hybrid quantum-classical algorithm. VQAs are an iterative approach to finding an approximate solution to a problem. They work hand in hand with classical optimisers to find solutions in a loop (Cerezo et al., 2021).
The loop that VQAs are built on starts with a parameterised quantum circuit (PQC), which prepares a state which is then measured. The measurement results are then fed into a classical optimiser (e.g. COBYLA, gradient descent), which produces an updated set of parameters for the circuit. The loop is iterated on until the results converge.
The parameters of a VQA can be thought of as knobs that can be adjusted and fine-tuned by the classical optimiser to modify the output of the quantum circuit until the results approximately address the problem.
The VQC used here
A VQC is a VQA used for classification. In this research each sample’s features are loaded onto the qubits (encoding), a trainable layer called the ansatz follows, and the class (malware or benign) is read from qubit 0.
VQA circuits are small enough to run on today’s quantum computers and small enough to be used in research simulations. VQAs are not a perfect solution: the noise that current quantum computers produce interferes with the parameter tuning portion of the loop, and problem sizes are restricted due to qubit counts, error rates and coherence times (Bharti et al., 2022).
Next: the qubit bottleneck