A Survey of Classical And Quantum Sequence Models

TL;DR

The paper surveys classical and quantum sequence models, focusing on RNNs and Transformer-based self-attention, and explores near-term quantum approaches (QRNN and QSANN) with practical enhancements. It re-implements representative quantum methods, adapts a quantum hybrid transformer to text and image tasks, and introduces encoding and positional strategies to boost performance. Key contributions include a detailed comparison between classical and quantum sequence models, experimental results on text and image classification, and evidence that positional encoding significantly improves QSANN convergence and accuracy. The work sheds light on how quantum self-attention might complement or exceed classical methods in specific contexts, while highlighting current hardware constraints and areas for further development to realize scalable quantum sequence models.

Abstract

Our primary objective is to conduct a brief survey of various classical and quantum neural net sequence models, which includes self-attention and recurrent neural networks, with a focus on recent quantum approaches proposed to work with near-term quantum devices, while exploring some basic enhancements for these quantum models. We re-implement a key representative set of these existing methods, adapting an image classification approach using quantum self-attention to create a quantum hybrid transformer that works for text and image classification, and applying quantum self-attention and quantum recurrent neural networks to natural language processing tasks. We also explore different encoding techniques and introduce positional encoding into quantum self-attention neural networks leading to improved accuracy and faster convergence in text and image classification experiments. This paper also performs a comparative analysis of classical self-attention models and their quantum counterparts, helping shed light on the differences in these models and their performance.

Figures (10)

• Figure 1: Data Encoding for QRNNs
• Figure 2: RNN state evolution
• Figure 3: Data Encoding for QSANNs
• Figure 4: Quantum Recurrent Block
• Figure 5: Ansatz Employed for QRNNs