Barren plateaus from learning scramblers with local cost functions

Abstract

The existence of barren plateaus has recently revealed new training challenges in quantum machine learning (QML). Uncovering the mechanisms behind barren plateaus is essential in understanding the scope of problems that QML can efficiently tackle. Barren plateaus have recently been shown to exist when learning global properties of random unitaries, which is relevant when learning black hole dynamics. Establishing whether local cost functions can circumvent these barren plateaus is pertinent if we hope to apply QML to quantum many-body systems. We prove a no-go theorem showing that local cost functions encounter barren plateaus in learning random unitary properties.

arXiv comment9 + 74 pages, 2 + 1 Figures

References in corpus (42)

• Supplementary information for "Quantum supremacy using a programmable superconducting processor"
• Quantum Machine Learning
• Probing many-body dynamics on a 51-atom quantum simulator
• A bound on chaos
• Barren plateaus in quantum neural network training landscapes
• Gapless Spin-Fluid Ground State in a Random Quantum Heisenberg Magnet
• Black holes and the butterfly effect
• Quantum Convolutional Neural Networks
• Black holes as mirrors: quantum information in random subsystems
• Cost Function Dependent Barren Plateaus in Shallow Parametrized Quantum Circuits
• Fast Scramblers
• Quantum many-body scars
• Noise-Induced Barren Plateaus in Variational Quantum Algorithms
• Power of data in quantum machine learning
• Integration with respect to the Haar measure on unitary, orthogonal and symplectic group
• Connecting ansatz expressibility to gradient magnitudes and barren plateaus
• Measuring the scrambling of quantum information
• Quantum advantage with shallow circuits
• An initialization strategy for addressing barren plateaus in parametrized quantum circuits
• Strong and weak thermalization of infinite non-integrable quantum systems
• Verified Quantum Information Scrambling
• Absence of Barren Plateaus in Quantum Convolutional Neural Networks
• Scrambling and thermalization in a diffusive quantum many-body system
• Unifying fast scrambling, thermalization and entanglement through the measurement of FOTOCs in the Dicke model
• Quantum speedup for active learning agents
• Trainability of Dissipative Perceptron-Based Quantum Neural Networks
• Large gradients via correlation in random parameterized quantum circuits
• Entanglement Devised Barren Plateau Mitigation
• Barren plateaus preclude learning scramblers
• Locality, Quantum Fluctuations, and Scrambling
• Exact emergent quantum state designs from quantum chaotic dynamics
• Accessing scrambling using matrix product operators
• Analyzing the barren plateau phenomenon in training quantum neural networks with the ZX-calculus
• Information Scrambling in Quantum Neural Networks
• Quantum scrambling with classical shadows
• The Presence and Absence of Barren Plateaus in Tensor-network Based Machine Learning
• Scrambling in the Dicke model
• Scrambling Ability of Quantum Neural Networks Architectures
• Emergent statistical mechanics from properties of disordered random matrix product states
• Chaos and Complexity from Quantum Neural Network: A study with Diffusion Metric in Machine Learning
• Hayden-Preskill decoding from noisy Hawking radiation
• Quantifying scrambling in quantum neural networks

Cited by in corpus (9)

• Barren Plateaus in Variational Quantum Computing
• A Lie Algebraic Theory of Barren Plateaus for Deep Parameterized Quantum Circuits
• Resource theory of quantum scrambling
• Constant-depth preparation of matrix product states with adaptive quantum circuits
• Absence of barren plateaus in finite local-depth circuits with long-range entanglement
• Isometric tensor network optimization for extensive Hamiltonians is free of barren plateaus
• Magic of Random Matrix Product States
• Absence of barren plateaus and scaling of gradients in the energy optimization of isometric tensor network states
• On the Hardness of Measuring Magic