Entanglement Induced Barren Plateaus

We argue that an excess in entanglement between the visible and hidden units in a Quantum Neural Network can hinder learning. In particular, we show that quantum neural networks that satisfy a volume-law in the entanglement entropy will give rise to models not suitable for learning with high probability. Using arguments from quantum thermodynamics, we then show that this volume law is typical and that there exists a barren plateau in the optimization landscape due to entanglement. More precisely, we show that for any bounded objective function on the visible layers, the Lipshitz constants of the expectation value of that objective function will scale inversely with the dimension of the hidden-subsystem with high probability. We show how this can cause both gradient descent and gradient-free methods to fail. We note that similar problems can happen with quantum Boltzmann machines, although stronger assumptions on the coupling between the hidden/visible subspaces are necessary. We highlight how pretraining such generative models may provide a way to navigate these barren plateaus.

[1]  Nathan Wiebe,et al.  Generative training of quantum Boltzmann machines with hidden units , 2019, ArXiv.

[2]  N. Wiebe,et al.  Tomography and generative training with quantum Boltzmann machines , 2016, 1612.05204.

[3]  Kristan Temme,et al.  Supervised learning with quantum-enhanced feature spaces , 2018, Nature.

[4]  Andreas Winter,et al.  Are random pure States useful for quantum computation? , 2008, Physical review letters.

[5]  Andrew T. Sornborger,et al.  Barren Plateaus Preclude Learning Scramblers. , 2020, Physical review letters.

[6]  Lu-Ming Duan,et al.  Machine learning meets quantum physics , 2019, Physics Today.

[7]  A. Harrow,et al.  Random Quantum Circuits are Approximate 2-designs , 2008, 0802.1919.

[8]  Seth Lloyd,et al.  Quantum-inspired low-rank stochastic regression with logarithmic dependence on the dimension , 2018, ArXiv.

[9]  Hartmut Neven,et al.  Classification with Quantum Neural Networks on Near Term Processors , 2018, 1802.06002.

[10]  Ryan Babbush,et al.  Barren plateaus in quantum neural network training landscapes , 2018, Nature Communications.

[11]  Maria Schuld,et al.  The quest for a Quantum Neural Network , 2014, Quantum Information Processing.

[12]  S. Lloyd,et al.  Quantum principal component analysis , 2013, Nature Physics.

[13]  M. Cerezo,et al.  Noise-induced barren plateaus in variational quantum algorithms , 2020, Nature Communications.

[14]  R. A. Low.  Large deviation bounds for k-designs , 2009, Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences.

[15]  Bin Yan,et al.  Information Scrambling and Loschmidt Echo. , 2019, Physical review letters.

[16]  Kunal Sharma,et al.  Trainability of Dissipative Perceptron-Based Quantum Neural Networks , 2020, ArXiv.

[17]  J. Eisert,et al.  Area laws for the entanglement entropy - a review , 2008, 0808.3773.

[18]  Teresa J. Feo,et al.  Structural absorption by barbule microstructures of super black bird of paradise feathers , 2018, Nature Communications.

[19]  A. Harrow,et al.  Approximate Unitary t-Designs by Short Random Quantum Circuits Using Nearest-Neighbor and Long-Range Gates , 2018, Communications in Mathematical Physics.

[20]  Matthew B. Hastings,et al.  Spectral Gap and Exponential Decay of Correlations , 2005 .

[21]  J. Eisert,et al.  Closing Gaps of a Quantum Advantage with Short-Time Hamiltonian Dynamics. , 2019, Physical review letters.

[22]  F. Haake.  Quantum signatures of chaos , 1991 .

[23]  F. Brandão,et al.  Local random quantum circuits are approximate polynomial-designs: numerical results , 2012, 1208.0692.

[24]  Masoud Mohseni,et al.  Quantum support vector machine for big feature and big data classification , 2013, Physical review letters.

[25]  Alán Aspuru-Guzik,et al.  On the Chemical Basis of Trotter-Suzuki Errors in Quantum Chemistry Simulation , 2014, 1410.8159.

[26]  A. J. Short,et al.  Entanglement and the foundations of statistical mechanics , 2005 .

[27]  Li Liu,et al.  Near-linear constructions of exact unitary 2-designs , 2015, Quantum Inf. Comput..

[28]  Roger Melko,et al.  Quantum Boltzmann Machine , 2016, 1601.02036.

[29]  D. Gross,et al.  Most quantum States are too entangled to be useful as computational resources. , 2008, Physical review letters.

[30]  Ewin Tang,et al.  Quantum-inspired classical algorithms for principal component analysis and supervised clustering , 2018, ArXiv.

[31]  Ashish Kapoor,et al.  Quantum deep learning , 2014, Quantum Inf. Comput..

[32]  Edward Grant,et al.  An initialization strategy for addressing barren plateaus in parametrized quantum circuits , 2019, Quantum.

[33]  Ewin Tang,et al.  A quantum-inspired classical algorithm for recommendation systems , 2018, Electron. Colloquium Comput. Complex..

[34]  Maria Schuld,et al.  Quantum Machine Learning in Feature Hilbert Spaces. , 2018, Physical review letters.

[35]  N. Wiebe.  Key questions for the quantum machine learner to ask themselves , 2020, New Journal of Physics.

[36]  Akira Sone,et al.  Cost-Function-Dependent Barren Plateaus in Shallow Quantum Neural Networks , 2020, ArXiv.