Knowledge Distillation in Quantum Neural Network using Approximate Synthesis
Knowledge Distillation in Quantum Neural Network using Approximate Synthesis
复制标题
DOI：
10.1145/3566097.3567877
复制
发表时间：
2022-07
期刊：
影响因子：
--
通讯作者：
M. Alam;Satwik Kundu;Swaroop Ghosh
中科院分区：
文献类型：
--
作者：
M. Alam;Satwik Kundu;Swaroop Ghosh
最近关于量子神经网络（QNN）在特定机器学习（ML）任务中的潜在优势的断言引发了相当数量的应用研究人员的好奇心。参数化量子电路（PQC）是QNN的主要构建模块，由几层单量子比特旋转和多量子比特纠缠操作组成。特定ML任务的PQC层的最佳数量通常是未知的。更大的网络通常在无噪声模拟中提供更好的性能。但是，与较浅的网络相比，它在硬件上的性能可能较差。由于量子器件之间的噪声量不同，PQC的最佳深度可能会有很大差异。此外，由于编译开销，为PQC选择的门可能适用于一种类型的硬件，但不适用于另一种类型的硬件。这使得难以将QNN设计推广到宽范围的硬件和噪声水平。另一种方法是针对每个硬件构建和训练多个QNN模型，这可能很昂贵。为了避免这些问题，我们引入了知识蒸馏的概念，QNN使用近似合成。所提出的方法将创建一个新的QNN网络，（i）减少层数或（ii）不同的门集，而无需从头开始训练。对新网络进行几个时期的训练，可以补偿逼近误差造成的损失。通过实证分析，我们证明了71.4%的电路层数减少，仍然达到了16.2%的噪声下更好的精度。
Recent assertions of a potential advantage of Quantum Neural Network (QNN) for specific Machine Learning (ML) tasks have sparked the curiosity of a sizable number of application researchers. The parameterized quantum circuit (PQC), a major building block of a QNN, consists of several layers of single-qubit rotations and multi-qubit entanglement operations. The optimum number of PQC layers for a particular ML task is generally unknown. A larger network often provides better performance in noiseless simulations. However, it may perform poorly on hardware compared to a shallower network. Because the amount of noise varies amongst quantum devices, the optimal depth of PQC can vary significantly. Additionally, the gates chosen for the PQC may be suitable for one type of hardware but not for another due to compilation overhead. This makes it difficult to generalize a QNN design to wide range of hardware and noise levels. An alternate approach is to build and train multiple QNN models targeted for each hardware which can be expensive. To circumvent these issues, we introduce the concept of knowledge distillation in QNN using approximate synthesis. The proposed approach will cre-ate a new QNN network with (i) a reduced number of layers or (ii) a different gate set without having to train it from scratch. Training the new network for a few epochs can compensate for the loss caused by approximation error. Through empirical analysis, we demonstrate ≈71.4% reduction in circuit layers, and still achieve ≈16.2% better accuracy under noise.