AAI_2025_Capstone_Chronicles_Combined

Breast Tumor Classification Using Quantum Neural Networks

6

Figure 3

Correlation matrix.

As quantum neural networks will be utilized for deep learning, there is feature mapping

that needs to be done. The classical (regular bits of the floating point values) values must be

mapped into quantum feature vectors. Specifically, this means taking every classical data point

x i and encoding it into the quantum vector |ψ(x i )> (Karthinkeyan, Akila, Sumathi, &

Poongodi, 2025). This encoding itself could be considered feature engineering. The largest

obstacle here is that there is a hard constraint of only being able to utilize approximately 20

simulated qubits. In other words, the data needs to be down sampled- at the expense of model

performance. This constraint also means that feature selection is performed, and the approach

259