Chi-Sheng Chen
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Chi-Sheng Chen has a total of 46 co-authors. Here are the top co-authors by number of shared publications:
- En-Jui Kuo
National yang Ming Chiao Tung University
4 shared publications - Shao-Hsuan Chang
Chang Gung University
4 shared publications Xinyu Zhang
3 shared publications- Che-Wei Liu
Department of Orthopaedics, Cathay General Hospital, Taipei 10630, Taiwan
2 shared publications Fan Zhang
2 shared publications- Guan-Ying Chen
Graduate Institute of Electronics Engineering, National Taiwan University, Taipei, Taiwan
2 shared publications
Latest Publications
- Open Access
- 29 July 2026
- Published by IOP Publishing in Quantum Science and Technology
- Vol. 11 (3) , 035051
Quantum Adaptive Self-Attention for quantum Transformer models
A recurring weakness in quantum machine learning (QML) is that reported ‘quantum advantages’ are seldom tested against a capacity-matched classical control, leaving it unclear whether a gain comes from the quantum substrate or from the architectural change that accompanies it. Our primary contribution is methodological: a protocol for attributing such gains honestly—a capacity-matched classical bottleneck of identical parameter budget, transparent reporting of where quantum does not help, and validation on real quantum hardware—which we develop and apply through a concrete case study. That case study is Quantum Adaptive Self-Attention (QASA), a hybrid Transformer that replaces the value projection of a single encoder layer with a 36-parameter parameterized quantum circuit (PQC), keeping all other layers classical. Across nine synthetic benchmarks and the real-world ETTh1 dataset, QASA improves on a full-capacity classical Transformer for chaotic and trend-dominated signals. To ask whether this is a genuinely quantum effect, we introduce a control rarely applied in QML—a capacity-matched classical bottleneck with the same parameter budget—and find that it matches the PQC on the error metrics. The gain is therefore attributable to the low-rank value-projection bottleneck (an architectural parsimony principle), not to quantumness; adding further quantum layers only degrades performance and trainability. We accordingly position the quantum layer not as a source of accuracy advantage but as a competitive instantiation of this principle: its low-rank compression onto the signal’s intrinsic dimensionality is matched by a classical bottleneck, so the gain is architectural rather than quantum. The quantum layer’s distinguishing features are physical—a high circuit entanglement (Meyer–Wallach Q = 0.981 with only 27 CNOTs, which we report as a circuit-level property) and, most concretely, noisy intermediate-scale quantum deployability, which we verify by executing the trained model on a real IBM Quantum processor (one-step prediction within 7.5 % of noiseless simulation on the quantum-favored task and 25 % , with overlapping error bars, on a classical-favored control; no error mitigation). Relative to other quantum sequence models (quantum long short-term memory (QLSTM), QnnFormer), QASA is moreover the most resource-efficient of the three: it reaches competitive accuracy with the fewest quantum parameters (36 vs 90–128) and the strongest entanglement-per-gate ( Q = 0.981 with 27 CNOTs vs QLSTM’s 56), and retains trainable gradients that deeper circuits lose. We argue that capacity-matched baselines and honest reporting of where quantum does not help are prerequisites for credible quantum-machine-learning claims.
- 29 July 2026
- Published by IOP Publishing in Quantum Science and Technology
- Vol. 11 (3) , 035051
Open AccessQuantum Adaptive Self-Attention for quantum Transformer models
A recurring weakness in quantum machine learning (QML) is that reported ‘quantum advantages’ are seldom tested against a capacity-matched classical control, leaving it unclear whether a gain comes from the quantum substrate or from the architectural change that accompanies it. Our primary contribution is methodological: a protocol for attributing such gains honestly—a capacity-matched classical bottleneck of identical parameter budget, transparent reporting of where quantum does not help, and validation on real quantum hardware—which we develop and apply through a concrete case study. That case study is Quantum Adaptive Self-Attention (QASA), a hybrid Transformer that replaces the value projection of a single encoder layer with a 36-parameter parameterized quantum circuit (PQC), keeping all other layers classical. Across nine synthetic benchmarks and the real-world ETTh1 dataset, QASA improves on a full-capacity classical Transformer for chaotic and trend-dominated signals. To ask whether this is a genuinely quantum effect, we introduce a control rarely applied in QML—a capacity-matched classical bottleneck with the same parameter budget—and find that it matches the PQC on the error metrics. The gain is therefore attributable to the low-rank value-projection bottleneck (an architectural parsimony principle), not to quantumness; adding further quantum layers only degrades performance and trainability. We accordingly position the quantum layer not as a source of accuracy advantage but as a competitive instantiation of this principle: its low-rank compression onto the signal’s intrinsic dimensionality is matched by a classical bottleneck, so the gain is architectural rather than quantum. The quantum layer’s distinguishing features are physical—a high circuit entanglement (Meyer–Wallach Q = 0.981 with only 27 CNOTs, which we report as a circuit-level property) and, most concretely, noisy intermediate-scale quantum deployability, which we verify by executing the trained model on a real IBM Quantum processor (one-step prediction within 7.5 % of noiseless simulation on the quantum-favored task and 25 % , with overlapping error bars, on a classical-favored control; no error mitigation). Relative to other quantum sequence models (quantum long short-term memory (QLSTM), QnnFormer), QASA is moreover the most resource-efficient of the three: it reaches competitive accuracy with the fewest quantum parameters (36 vs 90–128) and the strongest entanglement-per-gate ( Q = 0.981 with 27 CNOTs vs QLSTM’s 56), and retains trainable gradients that deeper circuits lose. We argue that capacity-matched baselines and honest reporting of where quantum does not help are prerequisites for credible quantum-machine-learning claims.
- Open Access
- 1 July 2026
- Published by IOP Publishing in Journal of Cosmology and Astroparticle Physics
- Vol. 2026 (07)
Accretion-driven squeezing of fuzzy dark matter halo cores in the Schrödinger-Newton framework
We investigate the impact of accretion onto supermassive black holes on the density profiles of a fuzzy dark matter soliton core at the center of a dark matter halo. Treating the supermassive black hole at the galactic center as a point mass, we numerically solve the Schrödinger-Newton equation for the scalar field. We find that the time-dependent perturbation induces a pronounced squeezing of the soliton profile, simultaneously shrinking the core size and increasing the central density. These results clarify how black hole growth can reshape fuzzy-dark-matter structures and may help reconcile tensions in observed galactic core properties.
- 1 July 2026
- Published by IOP Publishing in Journal of Cosmology and Astroparticle Physics
- Vol. 2026 (07)
Open AccessAccretion-driven squeezing of fuzzy dark matter halo cores in the Schrödinger-Newton framework
We investigate the impact of accretion onto supermassive black holes on the density profiles of a fuzzy dark matter soliton core at the center of a dark matter halo. Treating the supermassive black hole at the galactic center as a point mass, we numerically solve the Schrödinger-Newton equation for the scalar field. We find that the time-dependent perturbation induces a pronounced squeezing of the soliton profile, simultaneously shrinking the core size and increasing the central density. These results clarify how black hole growth can reshape fuzzy-dark-matter structures and may help reconcile tensions in observed galactic core properties.
- Open Access
- 15 June 2026
- Published by Elsevier in SSRN Electronic Journal
Quantum Temporal Convolutional Neural Networks for Cross-Sectional Equity Return Prediction: A Comparative Benchmark Study
We propose a quantum temporal convolutional neural network (QTCNN) for cross-sectional equity return prediction under a realistic stock ranking workflow. The model combines a classical temporal encoder with parameterefficient quantum convolution and pooling layers, then fuses quantum outputs with classical features for final prediction. Using the JPX Tokyo Stock Exchange benchmark, we compare QTCNN against strong classical and quantum baselines and evaluate performance under long-short portfolio metrics. Results indicate competitive predictive robustness and improved risk-adjusted performance across multiple experimental settings.
- 15 June 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessQuantum Temporal Convolutional Neural Networks for Cross-Sectional Equity Return Prediction: A Comparative Benchmark Study
We propose a quantum temporal convolutional neural network (QTCNN) for cross-sectional equity return prediction under a realistic stock ranking workflow. The model combines a classical temporal encoder with parameterefficient quantum convolution and pooling layers, then fuses quantum outputs with classical features for final prediction. Using the JPX Tokyo Stock Exchange benchmark, we compare QTCNN against strong classical and quantum baselines and evaluate performance under long-short portfolio metrics. Results indicate competitive predictive robustness and improved risk-adjusted performance across multiple experimental settings.
- Open Access
FreqLens: Interpretable Frequency Attribution for Time Series Forecasting
Time series forecasting models often lack interpretability, limiting their adoption in domains requiring explainable predictions. We propose \textsc{FreqLens}, an interpretable forecasting framework that discovers and attributes predictions to learnable frequency components. \textsc{FreqLens} introduces two key innovations: (1) \emph{learnable frequency discovery} -- frequency bases are parameterized via sigmoid mapping and learned from data with diversity regularization, enabling automatic discovery of dominant periodic patterns without domain knowledge; and (2) \emph{axiomatic frequency attribution} -- a theoretically grounded framework that provably satisfies Completeness, Faithfulness, Null-Frequency, and Symmetry axioms, with per-frequency attributions equivalent to Shapley values. On Traffic and Weather datasets, \textsc{FreqLens} achieves competitive or superior performance while discovering physically meaningful frequencies: all 5 independent runs discover the 24-hour daily cycle ($24.6 \pm 0.1$h, 2.5\% error) and 12-hour half-daily cycle ($11.8 \pm 0.1$h, 1.6\% error) on Traffic, and weekly cycles ($10\times$ longer than the input window) on Weather. These results demonstrate genuine frequency-level knowledge discovery with formal theoretical guarantees on attribution quality.
Open AccessFreqLens: Interpretable Frequency Attribution for Time Series Forecasting
Time series forecasting models often lack interpretability, limiting their adoption in domains requiring explainable predictions. We propose \textsc{FreqLens}, an interpretable forecasting framework that discovers and attributes predictions to learnable frequency components. \textsc{FreqLens} introduces two key innovations: (1) \emph{learnable frequency discovery} -- frequency bases are parameterized via sigmoid mapping and learned from data with diversity regularization, enabling automatic discovery of dominant periodic patterns without domain knowledge; and (2) \emph{axiomatic frequency attribution} -- a theoretically grounded framework that provably satisfies Completeness, Faithfulness, Null-Frequency, and Symmetry axioms, with per-frequency attributions equivalent to Shapley values. On Traffic and Weather datasets, \textsc{FreqLens} achieves competitive or superior performance while discovering physically meaningful frequencies: all 5 independent runs discover the 24-hour daily cycle ($24.6 \pm 0.1$h, 2.5\% error) and 12-hour half-daily cycle ($11.8 \pm 0.1$h, 1.6\% error) on Traffic, and weekly cycles ($10\times$ longer than the input window) on Weather. These results demonstrate genuine frequency-level knowledge discovery with formal theoretical guarantees on attribution quality.
- Open Access
A Unified SPD Token Transformer Framework for EEG Classification: Systematic Comparison of Geometric Embeddings
Spatial covariance matrices of EEG signals are Symmetric Positive Definite (SPD) and lie on a Riemannian manifold, yet the theoretical connection between embedding geometry and optimization dynamics remains unexplored. We provide a formal analysis linking embedding choice to gradient conditioning and numerical stability for SPD manifolds, establishing three theoretical results: (1) BWSPD's $\sqrtκ$ gradient conditioning (vs $κ$ for Log-Euclidean) via Daleckii-Kreĭn matrices provides better gradient conditioning on high-dimensional inputs ($d \geq 22$), with this advantage reducing on low-dimensional inputs ($d \leq 8$) where eigendecomposition overhead dominates; (2) Embedding-Space Batch Normalization (BN-Embed) approximates Riemannian normalization up to $O(\varepsilon^2)$ error, yielding $+26\%$ accuracy on 56-channel ERP data but negligible effect on 8-channel SSVEP data, matching the channel-count-dependent prediction; (3) bi-Lipschitz bounds prove BWSPD tokens preserve manifold distances with distortion governed solely by the condition ratio $κ$. We validate these predictions via a unified Transformer framework comparing BWSPD, Log-Euclidean, and Euclidean embeddings within identical architecture across 1,500+ runs on three EEG paradigms (motor imagery, ERP, SSVEP; 36 subjects). Our Log-Euclidean Transformer achieves state-of-the-art performance on all datasets, substantially outperforming classical Riemannian classifiers and recent SPD baselines, while BWSPD offers competitive accuracy with similar training time.Open AccessA Unified SPD Token Transformer Framework for EEG Classification: Systematic Comparison of Geometric Embeddings
Spatial covariance matrices of EEG signals are Symmetric Positive Definite (SPD) and lie on a Riemannian manifold, yet the theoretical connection between embedding geometry and optimization dynamics remains unexplored. We provide a formal analysis linking embedding choice to gradient conditioning and numerical stability for SPD manifolds, establishing three theoretical results: (1) BWSPD's $\sqrtκ$ gradient conditioning (vs $κ$ for Log-Euclidean) via Daleckii-Kreĭn matrices provides better gradient conditioning on high-dimensional inputs ($d \geq 22$), with this advantage reducing on low-dimensional inputs ($d \leq 8$) where eigendecomposition overhead dominates; (2) Embedding-Space Batch Normalization (BN-Embed) approximates Riemannian normalization up to $O(\varepsilon^2)$ error, yielding $+26\%$ accuracy on 56-channel ERP data but negligible effect on 8-channel SSVEP data, matching the channel-count-dependent prediction; (3) bi-Lipschitz bounds prove BWSPD tokens preserve manifold distances with distortion governed solely by the condition ratio $κ$. We validate these predictions via a unified Transformer framework comparing BWSPD, Log-Euclidean, and Euclidean embeddings within identical architecture across 1,500+ runs on three EEG paradigms (motor imagery, ERP, SSVEP; 36 subjects). Our Log-Euclidean Transformer achieves state-of-the-art performance on all datasets, substantially outperforming classical Riemannian classifiers and recent SPD baselines, while BWSPD offers competitive accuracy with similar training time. - Open Access
- 4 December 2025
- Published by Springer Nature in Journal of Signal Processing Systems
- Vol. 98 (1)
Exploring the Potential of QEEGNet for Cross-task and Cross-dataset Electroencephalography Encoding with Quantum Machine Learning
- 4 December 2025
- Published by Springer Nature in Journal of Signal Processing Systems
- Vol. 98 (1)
Open AccessExploring the Potential of QEEGNet for Cross-task and Cross-dataset Electroencephalography Encoding with Quantum Machine Learning
- 19 November 2025
- Published by Springer Nature in Medical & Biological Engineering & Computing
- Vol. 64 (2) , 701-714
Detection and tracking of a gauze sponge in minimally invasive surgery using a YOLO and R-CNN based model
Gauze sponges are the items most commonly retained from surgery. The additional time required to find the missing gauze sponge increases anesthetic risk and causes a delay for the next surgery. In minimally invasive surgery, a digital camera can record any object on the screen during surgical procedure. This study aimed to compare modern object detection methods and propose a gauze tracking model to detect and trace the location of gauze sponges in surgical videos. The model consisted of a detection module and a regulating module. The methods used in the detection module included the YOLO series and faster R-CNN with different backbones. The regulating module was designed to reduce false positive detections. The model detected gauze and converted an entire video into a timeline to illustrate segments when gauze appeared on the screen. The timeline was compared frame-by-frame with human annotations. Faster R-CNN, with ResNet101-FPN as the backbone, outperformed other methods. Adding a regulating module further increased the accuracy and F1-score to 0.94 and 0.862, respectively. The model was trained and tested using human surgical videos. The presence of gauze sponge identified by the model was consistent with human annotations. The results are promising for the possibility of real-time gauze tracking during surgery. The model is able to provide critical information to help surgeons locate missing gauze sponges. Graphical Abstract
- 19 November 2025
- Published by Springer Nature in Medical & Biological Engineering & Computing
- Vol. 64 (2) , 701-714
Detection and tracking of a gauze sponge in minimally invasive surgery using a YOLO and R-CNN based model
Gauze sponges are the items most commonly retained from surgery. The additional time required to find the missing gauze sponge increases anesthetic risk and causes a delay for the next surgery. In minimally invasive surgery, a digital camera can record any object on the screen during surgical procedure. This study aimed to compare modern object detection methods and propose a gauze tracking model to detect and trace the location of gauze sponges in surgical videos. The model consisted of a detection module and a regulating module. The methods used in the detection module included the YOLO series and faster R-CNN with different backbones. The regulating module was designed to reduce false positive detections. The model detected gauze and converted an entire video into a timeline to illustrate segments when gauze appeared on the screen. The timeline was compared frame-by-frame with human annotations. Faster R-CNN, with ResNet101-FPN as the backbone, outperformed other methods. Adding a regulating module further increased the accuracy and F1-score to 0.94 and 0.862, respectively. The model was trained and tested using human surgical videos. The presence of gauze sponge identified by the model was consistent with human annotations. The results are promising for the possibility of real-time gauze tracking during surgery. The model is able to provide critical information to help surgeons locate missing gauze sponges. Graphical Abstract
- Open Access
Quantum Adaptive Self-Attention for Financial Rebalancing: An Empirical Study on Automated Market Makers in Decentralized Finance
We formulate automated market maker (AMM) \emph{rebalancing} as a binary detection problem and study a hybrid quantum--classical self-attention block, \textbf{Quantum Adaptive Self-Attention (QASA)}. QASA constructs quantum queries/keys/values via variational quantum circuits (VQCs) and applies standard softmax attention over Pauli-$Z$ expectation vectors, yielding a drop-in attention module for financial time-series decision making. Using daily data for \textbf{BTCUSDC} over \textbf{Jan-2024--Jan-2025} with a 70/15/15 time-series split, we compare QASA against classical ensembles, a transformer, and pure quantum baselines under Return, Sharpe, and Max Drawdown. The \textbf{QASA-Sequence} variant attains the \emph{best single-model risk-adjusted performance} (\textbf{13.99\%} return; \textbf{Sharpe 1.76}), while hybrid models average \textbf{11.2\%} return (vs.\ 9.8\% classical; 4.4\% pure quantum), indicating a favorable performance--stability--cost trade-off.
Open AccessQuantum Adaptive Self-Attention for Financial Rebalancing: An Empirical Study on Automated Market Makers in Decentralized Finance
We formulate automated market maker (AMM) \emph{rebalancing} as a binary detection problem and study a hybrid quantum--classical self-attention block, \textbf{Quantum Adaptive Self-Attention (QASA)}. QASA constructs quantum queries/keys/values via variational quantum circuits (VQCs) and applies standard softmax attention over Pauli-$Z$ expectation vectors, yielding a drop-in attention module for financial time-series decision making. Using daily data for \textbf{BTCUSDC} over \textbf{Jan-2024--Jan-2025} with a 70/15/15 time-series split, we compare QASA against classical ensembles, a transformer, and pure quantum baselines under Return, Sharpe, and Max Drawdown. The \textbf{QASA-Sequence} variant attains the \emph{best single-model risk-adjusted performance} (\textbf{13.99\%} return; \textbf{Sharpe 1.76}), while hybrid models average \textbf{11.2\%} return (vs.\ 9.8\% classical; 4.4\% pure quantum), indicating a favorable performance--stability--cost trade-off.
- Open Access
Quantum Reinforcement Learning-Guided Diffusion Model for Image Synthesis via Hybrid Quantum-Classical Generative Model Architectures
Diffusion models typically employ static or heuristic classifier-free guidance (CFG) schedules, which often fail to adapt across timesteps and noise conditions. In this work, we introduce a quantum reinforcement learning (QRL) controller that dynamically adjusts CFG at each denoising step. The controller adopts a hybrid quantum--classical actor--critic architecture: a shallow variational quantum circuit (VQC) with ring entanglement generates policy features, which are mapped by a compact multilayer perceptron (MLP) into Gaussian actions over $Δ$CFG, while a classical critic estimates value functions. The policy is optimized using Proximal Policy Optimization (PPO) with Generalized Advantage Estimation (GAE), guided by a reward that balances classification confidence, perceptual improvement, and action regularization. Experiments on CIFAR-10 demonstrate that our QRL policy improves perceptual quality (LPIPS, PSNR, SSIM) while reducing parameter count compared to classical RL actors and fixed schedules. Ablation studies on qubit number and circuit depth reveal trade-offs between accuracy and efficiency, and extended evaluations confirm robust generation under long diffusion schedules.
Open AccessQuantum Reinforcement Learning-Guided Diffusion Model for Image Synthesis via Hybrid Quantum-Classical Generative Model Architectures
Diffusion models typically employ static or heuristic classifier-free guidance (CFG) schedules, which often fail to adapt across timesteps and noise conditions. In this work, we introduce a quantum reinforcement learning (QRL) controller that dynamically adjusts CFG at each denoising step. The controller adopts a hybrid quantum--classical actor--critic architecture: a shallow variational quantum circuit (VQC) with ring entanglement generates policy features, which are mapped by a compact multilayer perceptron (MLP) into Gaussian actions over $Δ$CFG, while a classical critic estimates value functions. The policy is optimized using Proximal Policy Optimization (PPO) with Generalized Advantage Estimation (GAE), guided by a reward that balances classification confidence, perceptual improvement, and action regularization. Experiments on CIFAR-10 demonstrate that our QRL policy improves perceptual quality (LPIPS, PSNR, SSIM) while reducing parameter count compared to classical RL actors and fixed schedules. Ablation studies on qubit number and circuit depth reveal trade-offs between accuracy and efficiency, and extended evaluations confirm robust generation under long diffusion schedules.
- Open Access
Quantum-Enhanced Natural Language Generation: A Multi-Model Framework with Hybrid Quantum-Classical Architectures
This paper presents a comprehensive evaluation of quantum text generation models against traditional Transformer/MLP architectures, addressing the growing interest in quantum computing applications for natural language processing. We conduct systematic experiments comparing five distinct models: Transformer (baseline), Quantum Kernel Self-Attention Network (QKSAN), Quantum RWKV (QRWKV), and Quantum Attention Sequence Architecture (QASA) across five diverse datasets including simple sentences, short stories, quantum phrases, haiku poetry, and proverbs. Our evaluation employs multiple metrics including perplexity, BLEU scores, vocabulary diversity, repetition rates, and fluency measures to assess different aspects of text generation quality. The experimental results reveal that while traditional Transformer models maintain overall superiority with the lowest average perplexity (1.21) and highest BLEU-1 score (0.2895), quantum-inspired models demonstrate competitive performance in specific scenarios. Notably, QKSAN achieves a competitive BLEU-1 score of 0.2800 while maintaining zero repetition rates, and QRWKV demonstrates perfect vocabulary diversity (Distinct-1 = 1.000) in certain tasks.
Open AccessQuantum-Enhanced Natural Language Generation: A Multi-Model Framework with Hybrid Quantum-Classical Architectures
This paper presents a comprehensive evaluation of quantum text generation models against traditional Transformer/MLP architectures, addressing the growing interest in quantum computing applications for natural language processing. We conduct systematic experiments comparing five distinct models: Transformer (baseline), Quantum Kernel Self-Attention Network (QKSAN), Quantum RWKV (QRWKV), and Quantum Attention Sequence Architecture (QASA) across five diverse datasets including simple sentences, short stories, quantum phrases, haiku poetry, and proverbs. Our evaluation employs multiple metrics including perplexity, BLEU scores, vocabulary diversity, repetition rates, and fluency measures to assess different aspects of text generation quality. The experimental results reveal that while traditional Transformer models maintain overall superiority with the lowest average perplexity (1.21) and highest BLEU-1 score (0.2895), quantum-inspired models demonstrate competitive performance in specific scenarios. Notably, QKSAN achieves a competitive BLEU-1 score of 0.2800 while maintaining zero repetition rates, and QRWKV demonstrates perfect vocabulary diversity (Distinct-1 = 1.000) in certain tasks.