.png){kind=icon}
.png){kind=icon}
Return
Quantum recurrent neural network for sequential labeling
.png){kind=icon}
PRE
AI
DOI：10.1007/s10115-026-02685-6.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Quantum computing is a computing paradigm that follows the laws of quantum mechanics, enabling the efficient solution of complex computational problems beyond the capabilities of classical systems. A combination of quantum computing and machine learning, known as quantum machine learning (QML), is a rapidly emerging field that has the potential to revolutionize traditional artificial intelligence workflows. Recently, QML algorithms have been applied to various natural language processing (NLP) applications, yielding promising results. However, little attention has been devoted to applying quantum-enhanced models to sequential labeling, a crucial task in NLP. To bridge this gap, this paper proposes a quantum recurrent neural network (QRNN) designed specifically for sequential labeling tasks. The proposed QRNN is a hybrid quantum-classical model, which combines quantum and classical resources. It utilizes data angle encoding as encoding methods to convert classical data into quantum states and variational layers to process quantum data. Comprehensive experiments are conducted on nine distinct code-mixed POS tagging datasets, covering multiple language pairs, to evaluate the performance and generalization of the QRNN. Additionally, the study includes ablation analyses that investigate the impact of various data encoding schemes, entanglement strategies, and depths of the variational layer on model performance. The experimental results demonstrate that the proposed QRNN consistently achieves higher tagging accuracy and fewer trainable parameters compared to both classical RNN models and state-of-the-art quantum long short-term memory (QLSTM).
Keywords:
Natural language processing
Machine learning
Quantum computing
Quantum machine learning
Quantum neural network
Journal
IF:
3.1
Papers:
532
Citations:
5.2K