40516 Quantum Machine Learning for Sensor Fusion
M. Arun Anoop, P. Karthikeyan
Combining machine learning with quantum computing provides a promising approach for image processing, medical imaging, and machine learning to ensure accuracy and speed. The utilization of quantum machine learning allows the same machine learning classifiers, which provide improved speed and performance accuracy rates for any kind of images or different modalities. The term “sensor fusion” refers to the integration of different imaging modalities or cross-domain features like spatial, frequency, metadata, and physical attributes to enhance detection capabilities. As we know, in classical computers, bits are the data storage units where values are either 0 or 1. In the case of quantum computers, bits are replaced by the term qubits or quantum bits. Qubits are managed by quantum computers, which are capable of solving more complex problems. In quantum machine learning (QML), modules like Qiskit and AerSimulators (which are quantum simulators) play a role in managing training and prediction before accuracy score generation. Primarily, QML works based on fidelity (similarity between two quantum states, meaning similarity between classical and quantum predictions), samplers, zzfeaturemap() (which converts classical features into a quantum-capable format), and adhoc kernels, along with classifiers before the training or fit function. Here, the sensor fusion we have considered involves categorical and numerical values from the Titanic survival dataset. In this chapter, we mainly concentrate on the phases required for sensor-used feature classification, and finally, primer evaluation will also be conducted with the help of the Titanic survival dataset to highlight the importance of the quantum machine learning branch in handling sensor fusion.
De Gruyter eBooks•Apr 28, 2025•0 citations