Cost-Function Locality and Trainability Diagnostics for Quantum Neural Networks: An Empirical and Artifact-Level Study of Layerwise and Local-Cost Training
We report an empirical study of how quantum neural networks (QNNs) train under global and local costfunctions, with layerwise training included as a third point of comparison. The main experiment is a set of 180 training runson binary MNIST (digits 3 and 6), using an 8-qubit hardware-efficient ansatz written in TensorFlow Quantum. A second,separate study measures gradient landscapes at a fixed input across 48 circuit configurations with 4 to 10 qubits. Wecompare three procedures: end-to-end training with a global product-Z readout (the baseline), a layerwise curriculum thatreuses trained parameters between stages, and training on a local averaged-Z cost. Test accuracy barely moves betweenmethods. The global baseline still reaches 87.2% at 8 layers, so we see no accuracy collapse in this regime, andmatched-seed tests at 8 layers find no significant difference (Friedman p = 0.654; every pairwise paired test isnon-significant, as are the tests at 4 and 6 layers). Gradient statistics tell the methods apart much more clearly. At depths 4,6 and 8, the local cost ends training with 1.5 to 2.4 times lower gradient variance than the global baseline (paired p <= 9.4x 10⁻³ after Holm correction; Cohen's d_z up to 2.3). It also costs far more to simulate at depth: 6.57 h per run at 8 layers,against 1.66 h for the baseline. In the landscape study, the local observable's gradient variance is 1.4 to 10 times lower thanthe global observable's over the whole configuration grid, and 4 to 10 times lower at depth 8. Layerwise training follows adifferent path, with gradient variance falling during optimization, yet it gives no
We report an empirical study of how quantum neural networks (QNNs) train under global and local costfunctions, with layerwise training included as a third point of comparison. The main experiment is a set of 180 training runson binary MNIST (digits 3 and 6), using an 8-qubit hardware-efficient ansatz written in TensorFlow Quantum. A second,separate study measures gradient landscapes at a fixed input across 48 circuit configurations with 4 to 10 qubits. Wecompare three procedures: end-to-end training with a global product-Z readout (the baseline), a layerwise curriculum thatreuses trained parameters between stages, and training on a local averaged-Z cost. Test accuracy barely moves betweenmethods. The global baseline still reaches 87.2% at 8 layers, so we see no accuracy collapse in this regime, andmatched-seed tests at 8 layers find no significant difference (Friedman p = 0.654; every pairwise paired test isnon-significant, as are the tests at 4 and 6 layers). Gradient statistics tell the methods apart much more clearly. At depths 4,6 and 8, the local cost ends training with 1.5 to 2.4 times lower gradient variance than the global baseline (paired p <= 9.4x 10⁻³ after Holm correction; Cohen's d_z up to 2.3). It also costs far more to simulate at depth: 6.57 h per run at 8 layers,against 1.66 h for the baseline. In the landscape study, the local observable's gradient variance is 1.4 to 10 times lower thanthe global observable's over the whole configuration grid, and 4 to 10 times lower at depth 8. Layerwise training follows adifferent path, with gradient variance falling during optimization, yet it gives no
This work presents a systematic empirical study of techniques for mitigating the barren plateau phenomena in quantum neural networks (QNNs). We conduct 45 controlled experiments for binary MNIST classification of digits 3 and 6 using a computational benchmark built using TensorFlow Quantum. We compare three different training procedures: a standard baseline setup, an incremental layerwise learning curriculum and a protocol based on local cost functions. For 8 layers we see mean accuracies of 73.9% and 75.4% for layerwise and local-cost interventions (over five stochastic seeds). The baseline setting drops from about 74% at 4 and 6 layers to 52.7% at 8 layers. Statistical reconstruction of the 8-layer regime suggests that the difference of 1.5 percentage points between layerwise and local-cost methods is not statistically significant, but both strategies are statistically significantly different from the baseline (one-way ANOVA F = 138.90; Welch tests p < 1.5 × 10−5). These results also suggest that practical training degradation is possible even when gradient magnitudes are above a conventional barren plateau detection threshold, highlighting the significance of continuous indicators for trainability, rather than binary threshold-based diagnostics.
Abstract Quantum neural networks (QNNs) are often challenged with the problem of flat cost function landscapes during training, known as barren plateaus (BP). A solution to potentially overcome the problem of the BP has recently been proposed by Cerezo et al In this solution, it is shown that, for an arbitrary deep quantum layer(s) in QNNs, a global cost function (all qubits measured in an n-qubit system) will always experience BP, whereas a local cost function (single qubit measured in an n-qubit system) can help to alleviate the problem of BP to a certain depth ( )). In this paper, we empirically analyze the locality and globality of the cost function in hybrid quantum neural networks. We consider two application scenarios namely, binary and multi-class classification, and show that for multiclass classification, the local cost function setting does not follow the claims of Cerezo et al; that is, the local cost function does not result in an extended quantum layer’s depth. We also show that for multiclass classification, the overall performance in terms of accuracy for the global cost function setting is significantly higher than the local cost function setting. On the other hand, for binary classification, our results show that the local cost function setting follows the claims of Cerezo et al, and results in an extended depth of quantum layers. However, the global cost function setting still performs slightly better than the local cost function.
Abstract With the increased focus on quantum circuit learning for near-term applications on quantum devices, in conjunction with unique challenges presented by cost function landscapes of parametrized quantum circuits, strategies for effective training are becoming increasingly important. In order to ameliorate some of these challenges, we investigate a layerwise learning strategy for parametrized quantum circuits. The circuit depth is incrementally grown during optimization, and only subsets of parameters are updated in each training step. We show that when considering sampling noise, this strategy can help avoid the problem of barren plateaus of the error surface due to the low depth of circuits, low number of parameters trained in one step, and larger magnitude of gradients compared to training the full circuit. These properties make our algorithm preferable for execution on noisy intermediate-scale quantum devices. We demonstrate our approach on an image-classification task on handwritten digits, and show that layerwise learning attains an 8% lower generalization error on average in comparison to standard learning schemes for training quantum circuits of the same size. Additionally, the percentage of runs that reach lower test errors is up to 40% larger compared to training the full circuit, which is susceptible to creeping onto a plateau during training.
Training quantum neural networks (QNNs) on near-term hardware remains hampered by two compounding difficulties: the exponential vanishing of gradient variance known as the barren plateau, and the O(L⋅2n) time and memory cost of differentiating through an n-qubit, L-layer circuit. We propose RLQ-Grad, a reinforcement-learning-based optimizer in which a classical policy πφ (a spectrally-normalized PPO agent) learns to propose parameter updates directly, conditioned on the QNN's current parameters, loss, accuracy, and previous update. Because the surrogate gradient is emitted by a classical network rather than obtained by differentiating through the unitary U(θ), its variance is not constrained by the barren plateau concentration bound, and its cost scales with the number of trainable parameters rather than the Hilbert-space dimension. We prove these properties formally and verify them on a hardware-efficient ansatz across four supervised benchmarks with up to n=20 qubits. RLQ-Grad preserves a near-flat gradient-variance curve where backpropagation, parameter-shift, and adjoint differentiation decay by 1 to 2 orders of magnitude. Accounting for the full training pipeline (PPO rollouts, actor-critic updates, and optimizer states), RLQ-Grad needs under 2 MB of memory and runs 2490×, 7876×, and 673× faster per iteration than these three methods at n=20. It improves top-1 accuracy by up to +10% over gradient-based baselines on circuits of up to 12 qubits, and matches dedicated barren plateau mitigation methods on CIFAR-10 at 14 to 20 qubits, where evolutionary and gradient-free optimizers collapse to chance.
The field of machine learning has experienced rapid development over the last decades and has found numerous applications across diverse domains ranging from industrial to personal settings. Quantum computing, on the other hand, is an emerging field offering a fundamentally different computational paradigm that can potentially surpass traditional computational approaches in certain domains.In this thesis, we explore the intersection between these two fields, commonly referred to as quantum machine learning (QML). In particular, we focus on Quantum neural networks (QNNs), which have gained attention in the field of QML due to the unparalleled success of their classical counterparts. However, the limitations of the current quantum computers have led to hybrid approaches combining classical and quantum devices. We, therefore, use hybrid quantum neural networks (HQNNs) to explore the potential advantages of QNNs in various practical applications.First, we explore the design space of HQNNs as there is no standardized methodology for designing quantum layers in HQNNs. Various quantum layer designs with different data encoding techniques are investigated, identifying the optimal set of quantum gates for different encoding strategies.We also investigate the impact of cost function globality and locality on the performance of HQNNs for real-world applications, namely: binary and multi-class classification. We exploit a single parameterized quantum circuit (PQC) structure with both single (local) and all-qubit (global) measurements. Our results show that the global cost function significantly outperforms the local cost function on multi-class classification. In contrast, both global and local cost functions have comparable performance on binary classification problems.Moreover, we propose a framework for analyzing the impact of various components of HQNNs with respect to each other on multi-class classification tasks, specifically focusing on the data encoding and quantum layers (expressibility and entanglement between qubits). Two ansatz structures, entangled and unentangled, are used to investigate the existence of the barren plateau (BP) in HQNNs. We further evaluate the impact of entanglement on trainability and provide recommendations for different constraint scenarios. The results show that the overall expressibility of the quantum layers impacts the presence of the BP phenomenon and that entanglement plays a role in the training landscapes of HQNNs, depending on the data encoding technique. Finally, we propose a new architecture for quantum neural networks (QNNs) called residual quantum neural networks (ResQNets), inspired by the classical residual neural networks (ResNets), to address the issue of BP in QNNs. ResQNets split the conventional QNN architecture into multiple quantum nodes (QNs), each with its own PQC, allowing residual connections between QNs. The results show that incorporating residual connections results in improved training performance, making ResQNets a promising solution for overcoming the problem of BP in QNNs.