On the interpretability of neural network decoders

Abstract

Neural-network (NN) based decoders are becoming increasingly popular in the field of quantum error correction (QEC), including for decoding of state-of-the-art quantum computation experiments. In this work, we make use of established interpretability methods from the field of machine learning, to introduce a toolbox to achieve an understanding of the underlying decoding logic of NN decoders, which have been trained but otherwise typically operate as black-box models. To illustrate the capabilities of the employed interpretability method, based on the Shapley value approximation, we provide an examplary case study of a NN decoder that is trained for flag-qubit based fault-tolerant (FT) QEC with the Steane code. We show how particular decoding decisions of the NN can be interpreted, and reveal how the NN learns to capture fundamental structures in the information gained from syndrome and flag qubit measurements, in order to come to a FT correction decision. Further, we show that the understanding of how the NN obtains a decoding decision can be used on the one hand to identify flawed processing of error syndrome information by the NN, resulting in decreased decoding performance, as well as for well-informed improvements of the NN architecture. The diagnostic capabilities of the interpretability method we present can help ensure successful application of machine learning for decoding of QEC protocols.

References in corpus (47)

• Surface codes: Towards practical large-scale quantum computation
• Methods for Interpreting and Understanding Deep Neural Networks
• Topological quantum memory
• Quantum Computing with Very Noisy Devices
• Suppressing quantum errors by scaling a surface code logical qubit
• Logical quantum processor based on reconfigurable atom arrays
• Topological Quantum Distillation
• Quantum error correction below the surface code threshold
• Demonstration of multi-qubit entanglement and algorithms on a programmable neutral atom quantum computer
• Exponential suppression of bit or phase flip errors with repetitive error correction
• Realization of an Error-Correcting Surface Code with Superconducting Qubits
• Active stabilisation, quantum computation and quantum state synthesis
• Demonstration of fault-tolerant universal quantum gate operations
• Low-distance Surface Codes under Realistic Quantum Noise
• Explaining a Series of Models by Propagating Shapley Values
• Fast Decoders for Topological Quantum Codes
• Experimental demonstration of fault-tolerant state preparation with superconducting qubits
• Efficient Algorithms for Maximum Likelihood Decoding in the Surface Code
• A Neural Decoder for Topological Codes
• Quantum Error Correction with the Toric-GKP Code
• Machine-learning-assisted correction of correlated qubit errors in a topological code
• Flag fault-tolerant error correction with arbitrary distance codes
• Decoding Small Surface Codes with Feedforward Neural Networks
• Encoding a magic state with beyond break-even fidelity
• Deep Neural Network Probabilistic Decoder for Stabilizer Codes
• Quantum error correction for the toric code using deep reinforcement learning
• Decoding color codes by projection onto surface codes
• Triangular color codes on trivalent graphs with flag qubits
• Deep neural decoders for near term fault-tolerant experiments
• Learning to Decode the Surface Code with a Recurrent, Transformer-Based Neural Network
• Parallel window decoding enables scalable fault tolerant quantum computation
• Comparing neural network based decoders for the surface code
• Neural network decoder for topological color codes with circuit level noise
• Demonstration of fault-tolerant Steane quantum error correction
• Deterministic correction of qubit loss
• Fault-tolerant quantum error correction for Steane's seven-qubit color code with few or no extra qubits
• A scalable and fast artificial neural network syndrome decoder for surface codes
• General framework for constructing fast and near-optimal machine-learning-based decoder of the topological stabilizer codes
• Efficient color code decoders in $d\ge 2$ dimensions from toric code decoders
• Convolutional neural network based decoders for surface codes
• A cellular automaton decoder for a noise-bias tailored color code
• Automated Synthesis of Fault-Tolerant State Preparation Circuits for Quantum Error Correction Codes
• Non-Markovian feedback for optimized quantum error correction
• Memory-assisted decoder for approximate Gottesman-Kitaev-Preskill codes
• Decoding quantum color codes with MaxSAT
• Neural Network-Based Design of Approximate Gottesman-Kitaev-Preskill Code
• Rescaling decoder for 2D topological quantum color codes on 4.8.8 lattices