Hybrid Classical-Quantum Machine Learning Algorithms for Large Scale Applications

In this talk, I will be discussing primarily on quantum machine learning, quantum optimization, hybrid classical-quantum neural networks with a direct application on computer vision, material science etc. Hybrid classical-quantum spiking and Random neural networks are highly promising candidate for quantum advantage. We propose a novel framework to demonstrate the feasibility of Hybrid Classical-Quantum Neural Networks (HCQNN) employing Variational Quantum Circuit (VQC) in the dressed quantum layer. The HCQNN relies on a hybrid classical-quantum circuit with gate parameters optimized during training. The dressed quantum layer in the suggested DSQ-Net model as a VQC are capable of being trained with thousands of parameters employed in the architecture. The HCQNN model has been experimented on various computer vision datasets using the Penny Lane quantum simulator.
Moreover, we are also working on Hybrid Parameterized Quantum Supervised Learning Classifiers. To obviate the data reduction before feeding to the circuit, dense parameterized quantum circuits (VQC) with lesser number trainable parameters have been proposed without compromising the classification accuracy. Recently, we have developed Quantum Kernel Integrated Ridge Regression for the direct applications to material science which will be also the part of our discussion. Finally, I will shed some light in to the future of quantum machine learning and the feasibility of quantum deep learning for large-scale applications.