Deep Learning – Deep Quantum Neural Networks

---

Quantum Machine Learning Simulator

Objective

To find a robust quantum neural network model by comparing the hybrid quantum-classical model according to quantum circuit design from a calibration perspective

Data

For quantum machine learning, we use Iris data provided by the UCI machine learning repository. The Iris data is classified into three categories according to the characteristics of the flower. The characteristics of the iris data are four in total

Related Work

Data embedded in Hilbert space are learned through parameterized quantum circuits. A total of 19 parameterized quantum circuits used in this study were made from a combination of gates to fairly compare expressiveness and entanglement capabilities between quantum circuits(Sim et al, 2019).

Proposed Method

For Quantum machine learning, data quantization is first performed through qubit embedding. Qubit embeddings use RX, RY, and RZ gates using minimal phenotypic embedding. The learning is then done through parameterized quantum circuits and the quantum qubit state is measured through Pauli Z gate in the measurement step. And finally, multi-classification is performed with Fully Connected layer.