AN OVERVIEW OF FOUNDATIONAL CONCEPTS,

ALGORITHMS & APPLICATIONS OF

QUANTUM MACHINE LEARNING

Mamta Rani1 , Ishaan Aditya2 , Arjun Beniwal3 , Sudarshan4 , and Deepanshu5

1 AssistantProfessor, Electrical and Electronics, Maharaja Surajmal Institute of Technology,
mamtatholia@msit.in
2 Electrical and Electronics,Maharaja Surajmal Institute Of Technology,

ishaanaditya108002@gmail.com
3 Electrical and Electronics, Maharaja Surajmal Institute of Technology,

arjunbeniwal07@gmail.com
4 Electrical and Electronics, Maharaja Surajmal Institute of Technology, sjagvir829@gmail.com
5 Electrical and Electronics, Maharaja Surajmal Institute of Technology,

ranadeepanshu91@gmail.com

Abstract. Quantum Machine Learning is an interdisciplinary combination of

quantum computing and classical machine learning. A review of fundamental
ideas, algorithms and real world uses in quantum machine learning is provided
in this paper. The foundations of quantum computing like Qubits, Quantum gates
and circuits are described first and then put in contrast with the main framework
and constraints of traditional machine learning. The study highlights important
methods like QSVM, Quantum k-means clustering and QPCA while classify-
ing algorithms into supervised and unsupervised. Also it investigates application
areas such as natural language processing, anomaly detection. The study also
highlights new approaches like Quantum PAC and Quantum Inspired algorithms.

Keywords: Quantum Machine Learning · Machine Learning · Natural Language

Processing · Optimization · Scalability

1 Introduction
Quantum Computing has changed computational science in areas where high dimen-
sional data processing and optimization are needed. Machine Learning, a subfield of
Artificial Intelligence has achieved success, but it struggles with scalability in large
datasets or NP-hard problems. For improvement of the learning model, Quantum Ma-
chine Learning uses the concepts of superposition, entanglement from quantum me-
chanics.
Combining data-driven intelligence with quantum hardware capabilities, QML sits
at the nexus of quantum physics, linear algebra, and algorithm design. The main distinc-
tions between classical and quantum computing are explained in this review, along with
important supervised and unsupervised quantum algorithms and their practical applica-
tions in fields like natural language processing, cybersecurity, and finance. The Paper’s
goal is to give an overview of Quantum Machine Learning’s present situation and future
possibilities, serving as an introduction for both novice and seasoned researchers.
2 Foundational Concepts

2.1 Fundamentals of Quantum Computing

Quantum Computing is an emerging field harnessing the aspects of Computer Sci-

ence, Physics & Mathematics. It utilizes quantum mechanics to solve complex prob-
lems faster than classical computers. As in Classical Computers classic bits exist in the
binary state (0,1) only whereas in Quantum Computers, Qubits exist in a linear combi-
nation of states which enables parallelism in computation. A quantum system’s progress
is governed by unitary transformations. Quantum Gates govern Qubits within Circuits
[1].
Qubit: Basic Information Unit of Quantum Computing, just like a bit in traditional
computing. A qubit can hold both states at the same time in superposition.
Quantum Gates: Operations in the circuit on qubits to perform logic. They are re-
versible and are unitary matrices. The gates correspond to operators used to manipulate
the quantum state of qubits. The only condition on a valid gate (and its corresponding
quantum operator) is that it should be unitary. By unitary we mean that it meets the
U † = U −1 condition, where U is the gate operation matrix and U † is the adjoint of U.
This property also ensures that it is a reversible gate operation [2].
Examples are: Hadamard Gate, Pauli-X Gate, CNOT Gate.

2.2 Classical Machine Learning

Fig. 1. Classical Machine Learning

It is a subset of Artificial Intelligence with focus on developing models to learn

from data without being explicitly programmed. It includes Supervised, Unsupervised
& Reinforcement learning.
Supervised learning includes model trained on labeled datasets to predict the out-
comes. Whereas in Unsupervised Learning, finding patterns or structures in unlabeled
data is needed. In Reinforcement learning feedback mechanism is used to train agents
to learn strategies through their interaction in various environments.
 Classical Machine Learning has it’s own limitations in computational scalability,
particularly with high dimensional data or NP-hard optimization problems. This leads
researchers to shift their focus to quantum approach [4].

2.3 Need for Quantum Machine Learning

The intersection of two main problems: (1) Classical ML algorithm’s growing resource
demand & (2) Exponential potential offered by Quantum Systems provides the need for
Quantum Machine Learning.
Quantum algorithmic approach allows more efficient operations on specific Ma-
chine Learning tasks. It can perform matrix inversion; Fourier Transforms and Sampling
tasks more efficiently than classical approach. Thus, for pattern recognition, optimiza-
tion and data classification quantum models are advised [1].

Fig. 2. Quantum Machine Learning

Table 1. Comparison between Machine Learning & Quantum Machine Learning

ASPECT MACHINE LEARNING QUANTUM MACHINE LEARN-

ING
Fundamental Unit Bits (0,1) only Qubits
Speed and Efficiency Limited by classical computational Potential for exponential speed for cer-
complexity. tain tasks.
Data Representation Numerical Vectors & Tensors. Quantum states.
Learning Algorithms Neural Networks, SVM, k-NN, deci- Quantum SVM’s, QNN, QPCA, Quan-
sion trees. tum Annealing.
Scalability Limited by memory & CPU/GPU Theoretically better for large scale
speed. problems, but current hardware limits
scalability.
3 Quantum Machine Learning Algorithms

Fig. 3. Quantum Mchine Learning Algorithm Use Case Proportion

3.1 Supervised Learning Algorithms

These algorithms rely on a quantum analog of a training dataset and have quantum cir-
cuits to approximate the mathematical functions or to discriminate data. They utilize
quantum resources to enhance classical models in tasks such as regression & classifica-
tion.
An example is Quantum Support Vector Machine (QSVM), which uses the quantum
kernel method and by encoding data in high dimensional Hilbert Spaces offers polyno-
mial speedup. It uses quantum inner product estimations for determination of margin
between data classes [13].
Quantum-enhanced nearest neighbor algorithm, where distance computations are
pushed through quantum amplitude estimations. For Quantum-native classification tasks,
quantum decision trees and perceptron models are considered [7].
Quantum linear regression, quantum k-NN and hybrid quantum-classical neural net-
works are quantum variants of supervised learning algorithms which use parametrized
quantum circuits(PQCs) to model complex decision boundaries while managing num-
ber of qubits [8].
Quantum annealing is another term used for supervised learning, particularly for
combinatorial optimization tasks such as feature selection and binary classification.
Quantum annealing frameworks (D-Wave Systems) can be mapped to Ising models or
Quadratic Unconstrained Binary Optimization (QUBO) problems, where the solution
represents the optimal classification configuration [9].
Supervised Methods (Quantum-Based) are applied to anomaly detection [6]. Quan-
tum distance-based classifier techniques and quantum enhanced decision trees are adapted
to identify rare or unusual data points which are promising in Cybersecurity and fraud
detection contexts.

3.2 Unsupervised Learning Algorithms

Clustering, dimensionality reduction and generative modeling are some of the unsuper-
vised learning algorithms that find hidden structures or patterns in unlabeled data.
Quantum k-means clustering is a method where distances between quantum-encoded
vectors are computed efficiently using quantum amplitude estimation and enables faster
convergence in partitioning large datasets [13].
Quantum Principal Component Analysis (QPCA) is an important technique which
uses the density matrix formalism to extract the principal components of a dataset en-
coded as quantum states. With the application of Unitary transformations and measure-
ments, QPCA identifies direction of maximal variance more efficiently than classical
PCA under certain conditions [13].
The integration of Quantum circuits for unsupervised tasks is still in progress but it
presents an exciting direction for large scale data mining [7].

Table 2. Comparison between Quantum Supervised & Unsupervised Learning

ASPECTS SUPERVISED LEARNING UNSUPERVISED LEARNING

Learning Type Interprets a map from inputs to known
Finds hidden pattern or structures in un-
outputs(labels). labeled data.
Algorithms QSVM, QNN, Quantum Decision Trees Quantum k-means, QPCA, Quantum
Clustering.
Data Requirements Labeled quantum encoded data. Works with unlabeled quantum en-
coded data.
Training Process Optimized over a cost function using la- Optimized over intra-data relationships
beled outcomes. and distances.
Complexity & Re- Depends on the size of labeled data and Depends on feature space representa-
sources circuit depth. tion and clustering algorithms.

4 Applications of Quantum Machine Learning

4.1 Financial Services

Portfolio Optimization, Credit-Risk Management and fraud detection are the areas where
quantum algorithms are extensively used. Quantum annealing methods solve optimiza-
tion problems that are common in financial modeling. QPCA can accelerate data pro-
cessing in large scale financial datasets while quantum enhanced classifiers support
anomaly detection in transactional data [13].
4.2 Pattern Recognition and Natural Language Processing (NLP)
Quantum algorithms can be used for text classification, document clustering and pattern
matching tasks. The algorithms have been explored to facilitate quantum enhanced NLP
applications. Quantum k-means and kernel based methods are also used for sentimental
analysis [13].

4.3 Cross-Domain Potential

Quantum Machine Learning has shown it’s potential in climate modeling, autonomous
vehicles and smart manufacturing [4]. The ability to process high velocity data streams
and discover patterns in complex environments with limited labeled data allows it to do
so.

Fig. 4. Application of Quantum Machine Learning

5 Emerging Direction
Quantum Machine Learning Theory explores situations where quantum algorithms out-
perform classical models focusing on sample complexity, generalization and oracle sep-
arations, helping identify problems where quantum models offer theoretical advantages
[10]. It forms the basis for exploring new learning areas such as Quantum PAC, statis-
tical and agnostic learning which may also influence artificial intelligence.

6 Conclusion
The study suggests that Quantum Machine Learning has the ability to solve significant
constraints in algorithmic complexity and data processing by integrating quantum prin-
ciples into classical Machine Learning Framework. The foundational concepts of quan-
tum machine learning were reviewed with supervised and unsupervised algorithms and
their uses in various areas. Quantum inspired algorithm approach was also noted for
further advancements. The future of Machine Learning is anticipated to be impacted by
Quantum Machine Learning as the theoretical research and quantum hardware develop.
References
1. Deutsch, D.: Quantum theory, the Church–Turing principle and the universal quantum com-
puter. Proceedings of the Royal Society of London. A. Mathematical and Physical Sciences
400(1818), 97–117 (1985)
2. Nielsen, M.A., Chuang, I.L.: Quantum Computation and Quantum Information, 10th An-
niversary Ed. Cambridge University Press, Cambridge (2010)
3. Rebentrost, P., Schuld, M., Wossnig, L., Petruccione, F., Lloyd, S.: Quantum gradient de-
scent and Newton’s method for constrained polynomial optimization. New Journal of Physics
21(7), 073023 (2019)
4. Garcı́a, D.P., Cruz-Benito, J., Garcı́a-Peñalvo, F.J.: Systematic literature review: Quantum
machine learning and its applications. arXiv preprint arXiv:2201.04093 (2022)
5. Schuld, M., Petruccione, F.: Supervised Learning with Quantum Computers. Springer Inter-
national Publishing, Cham (2018)
6. Corli, S., Moro, L., Dragoni, D., Dispenza, M., Prati, E.: Quantum machine learning algo-
rithms for anomaly detection: A review. arXiv preprint arXiv:2408.11047 (2024)
7. Abohashima, Z., Elhosen, M., Houssein, E.H., Mohamed, W.M.: Classification with quan-
tum machine learning: A survey. arXiv preprint arXiv:2006.12270 (2020)
8. Kulkarni, V., Kulkarni, M., Pant, A.: Quantum computing methods for supervised learning.
arXiv preprint arXiv:2006.12025 (2020)
9. Nath, R.K., Thapliyal, H., Humble, T.S.: A review of machine learning classification using
quantum annealing for real-world applications. arXiv preprint arXiv:2106.02964 (2021)
10. Arunachalam, S., de Wolf, R.: A survey of quantum learning theory. arXiv preprint
arXiv:1701.06806 (2017)
11. Benedetti, M., Lloyd, E., Sack, S., Fiorentini, M.: A quantum machine learning algorithm
based on generative models. Science Advances 5(10), eaau1946 (2019)
12. Bharti, K., Haug, T., Vedral, V., Kwek, L.-C.: Machine learning meets quantum foundations:
A brief survey. arXiv preprint arXiv:2003.11224 (2020)
13. Lloyd, S., Mohseni, M., Rebentrost, P.: Quantum algorithms for supervised and unsupervised
machine learning. arXiv preprint arXiv:1307.0411 (2013)