We explore the problem of barren plateaus [1] in Quantum Neural Networks: a certain large family of random quantum circuits have gradients that vanish almost everywhere.
Also, we explore the identity heuristic [2] initialization strategy as possible solution to overcome this problem.
This project follows the TensorFlow Quantum tutorial Barren plateaus [3].
- tensorflow 2.1.0
- tensorflow-quantum 0.3.1
Install them with:
$ pip install -r requirements.txtA quantum hardware is not required. The circuits are simulated using the Cirq library [4].
Run the experiments:
$ python3 barren.pyFor the oral exam of Quantum Information, I presented the experiments made alongside the theory from the papers using this presentation.
[1] J.R. McClean, S. Boixo, V.N. Smelyanskiy et al. Barren plateaus in quantum neural network training landscapes. (2018)
[2] E. Grant, L. Wossnig, M. Ostaszewski and M. Benedetti. An initialization strategy for addressing barren plateaus in parametrized quantum circuits. (2019)
[3] TensorFlow Quantum tutorials. Barren plateaus https://www.tensorflow.org/quantum/tutorials/barren_plateaus?hl=en
[4] Cirq library. https://quantumai.google/cirq
The directory cirq_examples contains some examples about using the Cirq library. They have been useful to understand how the library works.