Abstract:The potential impact of quantum machine learning algorithms on industrial applications remains an exciting open question. Conventional methods for encoding classical data into quantum computers are not only too costly for a potential quantum advantage in the algorithms but also severely limit the scale of feasible experiments on current hardware. Therefore, recent works, despite claiming the near-term suitability of their algorithms, do not provide experimental benchmarking on standard machine learning datasets. We attempt to solve the data encoding problem by improving a recently proposed variational algorithm [1] that approximately prepares the encoded data, using asymptotically shallow circuits that fit the native gate set and topology of currently available quantum computers. We apply the improved algorithm to encode the Fashion-MNIST dataset [2], which can be directly used in future empirical studies of quantum machine learning algorithms. We deploy simple quantum variational classifiers trained on the encoded dataset on a current quantum computer ibmq-kolkata [3] and achieve moderate accuracies, providing a proof of concept for the near-term usability of our data encoding method.

What problem does this paper attempt to address?

This paper mainly discusses the problem of using quantum machine learning algorithms on quantum computers to classify the Fashion-MNIST dataset. Currently, the methods of encoding classical data onto quantum computers are both expensive and limiting in terms of experimental scale. Therefore, despite some algorithms claiming to be suitable for near-term applications, there is a lack of benchmark testing on standard machine learning datasets. The paper proposes an improved variational algorithm that uses asymptotically shallow circuits to approximate the preparation of encoded data. These circuits adapt to the eigengate set and topology of the current quantum computer. Using this approach, they encoded the Fashion-MNIST dataset and made it publicly available for future research. After deploying a simple quantum variational classifier on the ibmq-kolkata quantum computer, a moderate level of accuracy was achieved, demonstrating the approximate practicality of the data encoding method. The main challenge faced in the study is the data encoding problem, i.e., how to effectively transform training and testing data into quantum states. Traditional methods require a trade-off between the number of qubits and the circuit depth. The algorithm proposed in the paper improves resource efficiency and reduces circuit depth, making it more faithful to implementing subsequent data processing tasks on existing quantum computers. The main contributions of the paper include: optimizing the parameterized quantum circuits (PQCs) in variational algorithms, reducing the circuit depth; encoding the Fashion-MNIST dataset in full scale using the improved algorithm; training and deploying a simple classifier on a quantum computer, achieving approximately 40% test accuracy, validating the feasibility of the data encoding method. In conclusion, this research aims to address the potential advantages of quantum machine learning in industrial applications and provides practical steps for benchmark testing of machine learning algorithms on quantum computers through an improved data encoding strategy.

Classification of the Fashion-MNIST Dataset on a Quantum Computer

Quantum Algorithms for Feedforward Neural Networks

Quantum-inspired classification based on quantum state discrimination

Empirical Power of Quantum Encoding Methods for Binary Classification

Supervised Learning Using a Dressed Quantum Network with "Super Compressed Encoding": Algorithm and Quantum-Hardware-Based Implementation

Accurate Image Multi-Class Classification Neural Network Model with Quantum Entanglement Approach

Nearest centroid classification on a trapped ion quantum computer

Quantum Hamiltonian Embedding of Images for Data Reuploading Classifiers

Quantum-Inspired Classification Algorithm from DBSCAN–Deutsch–Jozsa Support Vectors and Ising Prediction Model

Quantum variational distance-based centroid classifier

Quantum Neural Networks in Practice: A Comparative Study with Classical Models from Standard Data Sets to Industrial Images

A Quantum Computing-Based Accelerated Model for Image Classification Using a Parallel Pipeline Encoded Inception Module

Image Compression and Classification Using Qubits and Quantum Deep Learning

Leveraging Data Locality in Quantum Convolutional Classifiers

Using quantum transport networks for classification: A path toward quantum computing for machine learning

Quantum Transfer Learning for MNIST Classification Using a Hybrid Quantum-Classical Approach

Recent advances for quantum classifiers

Quantum machine learning for image classification

Quantum machine learning of large datasets using randomized measurements

Comparing Quantum Encoding Techniques

Experimental Realization of a Quantum Support Vector Machine.