---
title: 'Quanvolution: Quantum-Assisted Convolution'
url: https://www.emergentmind.com/topics/quanvolution
type: topic
---

# Quanvolution: Quantum-Assisted Convolution

Quanvolution, short for **quantum convolution**, denotes a family of hybrid quantum-classical operators in which local input patches are encoded into quantum states, processed by a quantum circuit, and measured to produce classical feature maps for downstream learning. In the dominant usage across quantum machine learning, quanvolution plays a role analogous to the first convolutional stage of a CNN, but it typically functions as a **quantum-assisted preprocessing stage** rather than as a fully quantum network. Locality is preserved, the same filter is applied repeatedly over patches, and the outputs are handed to classical components such as MLPs, CNN backends, Attention U-Nets, clustering pipelines, or AutoML systems [2407.17108] [2109.02862] [2507.13852].

## 1. Core concept and formal interpretations

A standard formalization writes the quanvolutional response for a local patch \(\mathbf{u}_x\) as
\[
f_x = d\big(q(e(\mathbf{u}_x))\big),
\]
where \(e(\cdot)\) is the encoding into a quantum state, \(q(\cdot)\) is the unitary quantum transformation, and \(d(\cdot)\) is the decoding stage, i.e. measurement and post-processing [2107.08171]. In this sense, quanvolution is the quantum analogue of a classical convolutional filter: it operates on local patches, uses multiple filters, and produces transformed feature maps, but the filter is a quantum circuit rather than a classical kernel.

Several papers emphasize that the analogy to classical convolution is structural rather than algebraic. Classical convolution computes local similarity through multiplication and summation. Quanvolution instead encodes each patch into a state such as \(\ket{\phi(x)}\), applies a circuit \(U_W(\vartheta)\), and derives features from expectation values of observables. One common expression is
\[
f(x,\vartheta) = \bra{\phi(x)}U^{\dagger}_{W}(\vartheta)\hat{O}U_W(\vartheta)\ket{\phi(x)},
\]
which makes the feature value an observable expectation after quantum evolution [2407.17108].

A more explicitly interpretable line of work reformulates quanvolution as quantum overlap estimation. In **Quantum Adjoint Convolution Operation (QACO)**, both a local patch and a kernel are amplitude-encoded, and their overlap is shown to be equivalent to the quantum normalization of convolution based on the **Frobenius inner product**. **Quantum Adjoint Convolutional Layer (QACL)** extends this primitive by **Quantum Phase Estimation (QPE)** so that all patch-wise Frobenius inner products can be computed in parallel [2404.17378]. This interpretation shifts quanvolution from a black-box variational circuit toward a similarity-based operator with a direct classical analogue.

The concept also appears in one-dimensional settings. In spectrum peak-finding, quanvolution is used as the quantum analogue of the convolutional feature-extraction stage in a 1D CNN: a local window is encoded, processed by a small PQC, and measured to produce output channels, while the rest of the network remains classical [2512.13125].

## 2. Patchwise pipeline and measurement semantics

Most quanvolutional systems implement a repeated patch pipeline. A local patch is extracted, encoded into a quantum state, transformed by a circuit, measured, and written back as a classical feature tensor. In the Tunis building-segmentation study, the workflow is explicit: **Sentinel-1 SAR GRD data** are acquired; the image is split into **\(256 \times 256\)** patches with **stride 128**; each patch is encoded into a quantum circuit using **angle encoding**; a quanvolutional operator is applied; the circuit output is measured to produce classical feature maps; and the resulting maps are fed to a classical **Attention U-Net** for **pixel-wise building segmentation** [2507.13852].

The embedding step is often written as
\[
U_{\phi}(\mathbf{x})\ket{0}^{\otimes n} = \ket{\phi(\mathbf{x})} = \ket{\psi},
\]
with a feature map \(\phi:\mathbb{R}^n \rightarrow H^{2^n}\). The patch values are therefore transformed into a state \(\ket{\psi}\) in a Hilbert space of dimension \(2^n\) [2507.13852]. In many implementations, measurement returns one scalar per qubit, and the resulting number of output features is directly tied to the number of qubits available in the circuit [2407.17108].

The measurement semantics vary. Some works use Pauli-\(Z\) expectation values on each qubit, arranged spatially to form a new image-like tensor [2109.02862]. Others decode with the expectation of \(Z^{\otimes n}\), or average repeated measurement outcomes into a scalar output [2410.05777]. In the 1D spectrum architecture, a **32-point** local window is encoded on **5 qubits**, one \(Z\)-expectation is measured per qubit, and each window therefore yields **5 channels**; with stride \(1\) over a **200-point** spectrum, the quantum convolution output has shape **\(5\times169\)** [2512.13125].

The role of the classical backend is correspondingly clear: quanvolution usually does not perform the final task by itself. In Earth Observation pipelines, quantum feature maps are handed to **BIRCH**, **AutoML**, or **AutoDL** [2407.17108]. In image classification studies, they are passed to an **MLP** or a fully connected classifier [2109.02862] [2106.07327]. In segmentation, they are placed **before** an Attention U-Net rather than replacing it [2507.13852].

## 3. Encoding strategies and circuit families

The design of a quanvolutional layer depends strongly on how classical patches are encoded into quantum states. Several distinct encoding families have been studied.

**Angle or rotational encoding** maps pixel values to qubit rotation angles. In the Tunis segmentation work, the input patch is represented using **angle encoding**, and a design constraint is stated as
\[
\text{qubits} \geq \text{kernel size}^2,
\]
ensuring enough qubits to represent the local patch [2507.13852]. In the rotational-encoding baseline examined elsewhere, a \(k\times k\) patch requires exactly \(k^2\) qubits, so the qubit count grows directly with kernel size [2410.05777].

**Threshold encoding**, **FRQI**, and **NEQR** represent a more explicit comparison among image encodings. Threshold encoding uses one qubit per pixel and stores only thresholded binary values; FRQI encodes intensity and position as
\[
|I\rangle = \frac{1}{2^n}\sum_{i=0}^{2^{2n}-1} (\cos \theta_i |0\rangle + \sin \theta_i |1\rangle)\otimes |i\rangle,
\]
using only \(2n+1\) qubits for a \(2^n \times 2^n\) image but with gate complexity \(\mathcal{O}(2^{4n})\); NEQR stores pixel values in basis states, uses \(8+2n\) qubits, and is dominated by multi-controlled operations [2106.07327]. These three encodings were compared specifically to determine how encoding interacts with variational quanvolution.

**Amplitude embedding** appears in the spectrum peak-finding architecture. There, a 32-point local window is normalized and embedded into the \(2^5=32\)-dimensional state space of a 5-qubit circuit, so the data vector becomes a superposition over the computational basis [2512.13125]. Amplitude encoding is also central to QACO/QACL, where both patch and kernel are encoded as amplitudes so that overlap reproduces the Frobenius inner product [2404.17378].

**Integrated encoding** is a more recent modification of the quanvolution pipeline. Instead of separating encoding and processing into two distinct circuit stages, the method merges them by injecting classical features directly into entangling gates of the form
\[
G(x_i)=\exp\big(-i\,\alpha(x_i)\,\sigma_1\otimes\sigma_2\big)[q_j,q_k].
\]
The tested angle mappings are **Simple**, **RndMul**, and **RndLin**, and the number of gates \(L\) is chosen to satisfy \(L\ge k^2\), with experiments using \(L=2k^2\) [2410.05777]. This removes the hard link between patch size and qubit count: the integrated model uses only **\(n=4\) qubits** even for filters as large as **\(5\times5\)**, whereas rotational encoding would require **25 qubits** for that patch size [2410.05777].

Circuit families also vary. The Tunis segmentation study compares **Strongly Entangled Circuit**, **Random Circuit**, and **Basic Entangled Circuit**, all with **9 qubits**, **2 layers**, **\(3\times3\) kernel**, and **9 filters** [2507.13852]. Another image-classification framework uses repeated **CRZ** entangling gates followed by **RY** rotations, taken as “Circuit 13” from Sim et al. [2109.02862]. The spectrum peak-finding work evaluates a **strongly entangling ansatz**, a **simplified Two-Design ansatz**, and a **random ansatz** in which all rotation gates are trainable [2512.13125].

## 4. Training regimes, parameterization, and resource trade-offs

A major divide in the literature concerns whether quantum filters are fixed, unsupervisedly selected, or fully trainable.

Many NISQ-oriented systems adopt **frozen quantum kernels**. Quanv4EO explicitly uses a **lazy training regime** in which quantum parameters are not trained and only classical stages are optimized [2407.17108]. The Tunis segmentation model also treats quanvolution as a fixed feature extractor placed before a classical Attention U-Net [2507.13852]. This regime is motivated by reduced computational burden, avoidance of difficult quantum optimization, and better compatibility with shallow circuits on limited hardware.

A more structured alternative is **unsupervised filter selection**. In the fault-detection study, candidate circuit ansätze are measured repeatedly, their output distributions are embedded as vectors in \(\mathbb{R}^{2^n}\), and **K-means clustering** is used to choose circuits near cluster centers. The stated goal is to maximize the difference of quantum properties such as expressibility and entangling capability, so that the final quanvolution filters are representative and diverse. Once the filters are fixed, features are extracted for the **entire training set just once**, and the downstream classifier is trained classically [2107.08171].

By contrast, several papers argue for **trainable quanvolution**. One framework makes a case for trainable quantum filters and reports that, after **10 epochs**, trainable filters achieved on average **15.98% lower training loss**, **7.49% lower validation loss**, **3.46% higher training accuracy**, and **3.32% higher validation accuracy** than non-trainable filters, although training required about **195 seconds per epoch** versus **57 seconds per epoch** for non-trainable filters on a Core i7-10750H machine with 16 GB RAM [2109.02862]. Another study shows that trainability interacts strongly with encoding choice: for **FRQI** with a **\(2\times2\)** filter, test accuracy improved from **0.806** to **0.854**, whereas for **Threshold encoding** with a **\(4\times4\)** filter, the untrainable model at **0.828** exceeded the trainable one at **0.802** [2106.07327]. In spectrum peak-finding, trainable quanvolution layers improved final training loss by about **15–25%**, but at higher computational cost [2512.13125].

Parameter and execution efficiency are therefore central. In building segmentation over Tunis, **Standard Attention U-Net** has **34.8 million trainable parameters**, whereas **Quanvolution + Attention U-Net** has **2.1 million trainable parameters**, a reduction of **over 93%**; the trade-off is a drop in **OA** from **0.9491** to **0.9343–0.9384**, with **Basic Entangled Circuit** performing best at **0.9384** [2507.13852]. Quanv4EO reports **42k parameters + 16 frozen qubits** on both MNIST and Fashion MNIST [2407.17108].

Execution cost can also be reduced at preprocessing time. Flexible quantization with memoization maps pixel values into an arbitrary number of levels \(N\), stores previously computed outputs for identical quantized patches, and reduces circuit executions to the number of unique quantized patches actually present in the dataset. For the main classification experiments on MiraBest and LArTPC, **\(N=50\)** yielded execution reductions of about **96%** and **97%**, respectively, while keeping **MSE below \(10^{-5}\)** [2410.05777].

## 5. Architectural compositions and application domains

Quanvolution has been applied across classification, segmentation, filtering, and signal-analysis settings, but almost always as the first feature-extraction stage in a hybrid architecture.

In image classification, a common pattern is a single quanvolutional layer followed by a classical classifier. One framework uses a quanvolutional layer followed by two fully connected classical layers [2109.02862]. Another uses a quanvolutional layer and a classical fully connected layer trained with cross-entropy on MNIST, comparing Threshold, FRQI, and NEQR encodings [2106.07327]. A more hierarchical design stacks multiple quanvolutional layers with **max-pooling** at each level and ends with a **three-layer fully connected neural network** [2107.08171].

In Earth Observation, quanvolution is used both for standard benchmark datasets and for remote-sensing tasks. Quanv4EO reports **accuracy = 0.9984** on MNIST and **96.81%** on Fashion MNIST with **42k + 16 frozen qubits**. On **EuroSAT**, quanvolutional features improve mean accuracy from **0.94** to **0.96** under **BIRCH** clustering, from **0.89** to **0.91** with **AutoSklearn**, and from **0.88** to **0.93** with **AutoKeras**. The same work also reports **Sentinel-1 SAR** speckle-filtering improvements from **PSNR 19.21 → 21.72** and **SSIM 0.75 → 0.81** [2407.17108].

In SAR building segmentation over Tunis, quanvolution is explicitly inserted **before** a classical **Attention U-Net**. The motivation is that dense urban SAR scenes are structurally complex, noisy, and high dimensional, and the claimed benefit is more informative feature maps that capture structural details while reducing the burden on the downstream network. The reported outcome is a strong accuracy-efficiency trade-off rather than absolute superiority over the standard model [2507.13852].

In classification tasks designed to probe resource usage, integrated encoding changes the comparison with both classical CNNs and rotational quanvolution. On **MiraBest**, the CNN baseline achieves about **71.0%** accuracy, **Rot** peaks around **77.3%** at \(k=2\), and **Int** with **Simple** or **RndMul** reaches about **80.8%** at \(k=3\). On **LArTPC**, the CNN reaches about **56.8%**, **Rot** peaks near **52.9%**, and **Int** with **Simple** or **RndMul** reaches about **58.5%** at \(k=4\) [2410.05777].

In one-dimensional signal analysis, quanvolution has been used for **bearing fault detection** and **spectrum peak-finding**. The fault-detection study uses **299 samples**, each a **\(192\times1\)** time series, with a **two-level hierarchy** of learned quantum feature extractors and reports competitive results [2107.08171]. The spectrum study uses synthetic NMR-inspired spectra and reports that, on the dedicated hard dataset, the strongly entangling ansatz yields roughly **10.9% improvement in F1** for peak counting and **29.8% lower MAE** for peak position estimation relative to the classical CNN; the abstract summarizes the result as an **11% improvement in F1 score** and a **30% reduction in mean absolute error** [2512.13125].

## 6. Conceptual debates, misconceptions, and adjacent usages

A recurrent misconception is that quanvolution is itself the final predictive model. In most of the cited work, this is incorrect: quanvolution is a **front-end preprocessing layer** or **feature-extraction layer**, while classification, clustering, segmentation, or regression remains classical [2507.13852] [2407.17108]. A second misconception is that quanvolution necessarily means **fixed random quantum filters**. The literature now includes frozen kernels, unsupervisedly selected filter banks, and end-to-end trainable quantum filters, so “quanvolution” names a broader design space rather than a single training regime [2107.08171] [2109.02862] [2512.13125].

Another important point is that reported gains are conditional rather than universal. Several papers present quanvolution as a favorable **accuracy-efficiency trade-off**, not as a uniformly dominant replacement for classical convolution. Standard Attention U-Net still has the highest reported **OA = 0.9491** in the Tunis study, exceeding all quantum-assisted variants [2507.13852]. In the image-encoding comparison, the best **\(4\times4\)** result is **Threshold untrainable** rather than a trainable model [2106.07327]. In the integrated-encoding study, the **Rot** baseline remains below the CNN on LArTPC, while the **Int** model exceeds it only for some settings [2410.05777]. In spectrum peak-finding, the authors explicitly avoid claiming hardware-level quantum advantage, and noisy simulations remain worse than ideal ones [2512.13125].

Interpretability and mathematical grounding are also active themes. QACO/QACL argue that many quantum convolutional layers are black-box structures and replace them with a Frobenius-inner-product interpretation based on amplitude encoding and overlap estimation [2404.17378]. A plausible implication is that part of the field’s development is moving from heuristic circuit feature maps toward more explicitly analyzable operators.

Two adjacent usages further delimit the term. **Monarq** is conceptually close to quanvolution because it performs front-end quantum data processing for convolution, DFT, squared gradients, and edge detection, but it is **not mainly a variational feature extractor**; it is a **non-variational, deterministic transform engine** built from **QCrank** and **EHands**, and the paper explicitly states that it is not “quanvolution” in the usual ML sense [2603.03582]. By contrast, **QUAN** uses “quanvolution” to refer to **quaternion convolution**, not quantum convolution: RGB inputs are embedded into quaternion space and processed by an approximate Hamilton-product decomposition implemented with real-valued convolutions [2509.05512]. For technical writing, this distinction is essential.

Taken together, the literature presents quanvolution as a patchwise quantum feature-extraction paradigm that remains experimentally heterogeneous. Its central variables are the encoding map, circuit family, readout rule, and training regime; its strongest empirical cases tend to involve difficult local structure, limited parameter budgets, or NISQ-motivated resource constraints; and its most robust claims concern representational flexibility and architecture-level trade-offs rather than universal accuracy gains.

Source: https://www.emergentmind.com/topics/quanvolution