This project aims to compare the performance of a quantum-enhanced feature space model using Qiskit on a real quantum backend with 100 qubits, against a classical neural network model running on a GPU using TensorFlow and Keras. The classical model is a simple neural network with two hidden layers (64 and 32 neurons).
The quantum-enhanced feature space model leverages quantum computing to map data into a high-dimensional space using quantum circuits, potentially capturing complex patterns that classical models might miss. By utilizing 100 qubits, this approach explores the capabilities of quantum processors to handle more intricate feature representations. In contrast, the classical neural network processes the original features of the dataset through its layers, relying on the power of GPU acceleration for efficient computation.
This code includes steps to save and load your IBM Quantum account, prepare the Iris dataset, train and evaluate both models, and visualize the results.
