Theoretical Guarantees for Permutation-Equivariant Quantum Neural Networks
read the original abstract
Despite the great promise of quantum machine learning models, there are several challenges one must overcome before unlocking their full potential. For instance, models based on quantum neural networks (QNNs) can suffer from excessive local minima and barren plateaus in their training landscapes. Recently, the nascent field of geometric quantum machine learning (GQML) has emerged as a potential solution to some of those issues. The key insight of GQML is that one should design architectures, such as equivariant QNNs, encoding the symmetries of the problem at hand. Here, we focus on problems with permutation symmetry (i.e., the group of symmetry $S_n$), and show how to build $S_n$-equivariant QNNs. We provide an analytical study of their performance, proving that they do not suffer from barren plateaus, quickly reach overparametrization, and generalize well from small amounts of data. To verify our results, we perform numerical simulations for a graph state classification task. Our work provides the first theoretical guarantees for equivariant QNNs, thus indicating the extreme power and potential of GQML.
This paper has not been read by Pith yet.
Forward citations
Cited by 5 Pith papers
-
Measurement-based quantum machine learning
The authors introduce MuTA as a universal quantum neural network for MBQC and numerically demonstrate its ability to learn gates, classify quantum states, and process data under noise, including photonic hardware constraints.
-
Resource-efficient equivariant quantum convolutional neural networks
Equivariant sp-QCNN encodes general symmetries with group theory, splits circuits at pooling layers to preserve symmetry while enabling parallel measurements, and shows improved efficiency and trainability over standa...
-
The power and limitations of learning quantum dynamics incoherently
The paper proves sample complexity bounds showing that any efficiently representable unitary can be learned incoherently with arbitrary measurements, but only low-entangling unitaries with shallow-depth measurements, ...
-
Exploiting More Than Symmetry in Variational Quantum Machine Learning
Suitable subgroups preserve most generalization benefit in equivariant VQCs, but dominant gains come from placing trainable gates on task-decisive motifs rather than from varying symmetry level.
-
Symmetries and overparametrization properties of Hamiltonian variational ansatzes for the $(1+1)$d $\mathbb{Z}_2$ lattice gauge theory
Numerical study of five symmetry-preserving HVAs for Z2 gauge theory finds overparametrization eliminates local minima and loss decay rate scales linearly with number of parameters.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.