Understanding Quantum Machine Learning With Qiskit

Introduction

Quantum Computing is an emerging field that incorporates quantum mechanics principles into computing. With Quantum Machine Learning (QML), we blend quantum computing and machine learning to create more efficient algorithms.

IBM’s quantum computing framework, Qiskit, allows us to explore QML. A key component to know in QML is Quantum Support Vector Machines (QSVM).

This post addresses QSVM implementation using Qiskit for binary classification problems. For better understanding, familiarity with basic machine learning and quantum computing is beneficial.

QSVM

Quantum Support Vector Machines use quantum states to represent data, which increases computational speed, providing a unique approach for solving complex machine learning problems.

from qiskit import BasicAer
from qiskit.ml.datasets import ad_hoc_data
from qiskit.aqua.utils import split_dataset_to_data_and_labels, map_label_to_class_name
from qiskit.aqua import QuantumInstance
from qiskit.aqua.algorithms import QSVM, SklearnSVM

Here, we import necessary components from Qiskit.

Data

For simplicity, we use ad_hoc_data from Qiskit's built-in datasets.

feature_dim = 2 # dimension of each data point
training_dataset_size = 20
testing_dataset_size = 10
random_seed = 10598

sample_Total, training_input, test_input, class_labels = ad_hoc_data(
    training_size=training_dataset_size, 
    test_size=testing_dataset_size, 
    n=feature_dim, gap=0.3, plot_data=True
)
datapoints, class_to_label = split_dataset_to_data_and_labels(test_input)

This loads the data with 20 training samples and 10 testing samples.

QSVM Execution

After setting up our dataset, we run the QSVM.

seed = 10598

backend = BasicAer.get_backend('qasm_simulator')
feature_map = SecondOrderExpansion(feature_dimension=feature_dim, depth=2, entanglement='linear')
svm = QSVM(feature_map, training_input, test_input)
quantum_instance = QuantumInstance(backend, shots=1024, seed=seed, seed_transpiler=seed)

result = svm.run(quantum_instance)

Here, we are using 'qasm_simulator' backend of Qiskit's BasicAer.

Results

Finally, we can examine the results.

print(f'Testing success ratio: {result['testing_accuracy']}') 

Conclusion

This rudimentary introduction to QSVM with Qiskit gives an initial idea of how Quantum Machine Learning can boost conventional machine learning algorithms. The power of quantum states representation is best seen with large complex problems, which can exponentially enhance efficiency.