Guan-Ying Chen
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Guan-Ying Chen has a total of 20 co-authors. Here are the top co-authors by number of shared publications:
- Chih-Ting Lin
Graduate Institute of Electronics Engineering, National Taiwan University, Taipei, Taiwan
4 shared publications - Chi-Sheng Chen
Graduate Institute of Biomedical Electronics and Bioinformatics, National Taiwan University, No. 1, Sec. 4, Roosevelt Road, Taipei 10617
2 shared publications Qiuzhe Xie
2 shared publicationsCheng-Yu Lin
National Taiwan University
1 shared publication- Chia-Fen Chu1 shared publication
Chih-Ting Lin
1 shared publication
Latest Publications
- 13 February 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Sensors Journal
- Vol. 26 (7) , 11342-11350
Deep Learning-Assisted Field-Effect Transistor for Polychromatic Light Sensing and Recognition
!CMOS image sensors have been widely utilized for image capture; however, they rely on color filtering and demosaicing algorithms for color reconstruction in each pixel. This single-photodetecter, single-color architecture constrains light utilization and image resolution. In this study, we proposed a deep learning (DL)-assisted field-effect transistor (FET) to decouple mixed light components into their respective wavelengths and intensities simultaneously. Through sequential-bilateral-voltage driving, a single FET generates a series of drain current shifts (DCSs) driven by transient photoelectric effects. Using a convolutional neural network (CNN), the DCS map is decoded into multiple light components. Experiments were conducted with combinations of wavelengths—635 nm (λred), 510 nm (λgreen), and 450 nm (λblue)—and intensities ranging from 0.1 to 0.9 W/cm². In monochromatic light experiments, the DCS-CNN achieved an average mean squared error (MSE) of 0.0014 and mean absolute error (MAE) of 0.0216, outperforming the baseline flat-DCS multilayer perceptron (MLP) by 86% in MSE and 43% in MAE, respectively. Additionally, our experiments confirmed the robustness of the DCS training method under laser source conditions, with and without 90-degree rotation. In polychromatic light experiments, the proposed DCS-CNN achieved 84.5% accuracy detecting light from 64 distinct combinations. To deepen model understanding, we investigated the impact of transistor operation regions and transient current variation maps on light detection capabilities. Overall, this study demonstrates the potential of the DL-FET architecture for enabling single-shot color sampling in cameras without color filters.
- 13 February 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Sensors Journal
- Vol. 26 (7) , 11342-11350
Deep Learning-Assisted Field-Effect Transistor for Polychromatic Light Sensing and Recognition
!CMOS image sensors have been widely utilized for image capture; however, they rely on color filtering and demosaicing algorithms for color reconstruction in each pixel. This single-photodetecter, single-color architecture constrains light utilization and image resolution. In this study, we proposed a deep learning (DL)-assisted field-effect transistor (FET) to decouple mixed light components into their respective wavelengths and intensities simultaneously. Through sequential-bilateral-voltage driving, a single FET generates a series of drain current shifts (DCSs) driven by transient photoelectric effects. Using a convolutional neural network (CNN), the DCS map is decoded into multiple light components. Experiments were conducted with combinations of wavelengths—635 nm (λred), 510 nm (λgreen), and 450 nm (λblue)—and intensities ranging from 0.1 to 0.9 W/cm². In monochromatic light experiments, the DCS-CNN achieved an average mean squared error (MSE) of 0.0014 and mean absolute error (MAE) of 0.0216, outperforming the baseline flat-DCS multilayer perceptron (MLP) by 86% in MSE and 43% in MAE, respectively. Additionally, our experiments confirmed the robustness of the DCS training method under laser source conditions, with and without 90-degree rotation. In polychromatic light experiments, the proposed DCS-CNN achieved 84.5% accuracy detecting light from 64 distinct combinations. To deepen model understanding, we investigated the impact of transistor operation regions and transient current variation maps on light detection capabilities. Overall, this study demonstrates the potential of the DL-FET architecture for enabling single-shot color sampling in cameras without color filters.
- 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.
- 8 December 2025
- Published by Elsevier in Sensors and Actuators B: Chemical
- Vol. 450
An effective strategy for aminosilane functionalization on dielectric surfaces in advanced biosensing applications
- 8 December 2025
- Published by Elsevier in Sensors and Actuators B: Chemical
- Vol. 450
An effective strategy for aminosilane functionalization on dielectric surfaces in advanced biosensing applications
- Open Access
- 18 May 2025
- Published by Elsevier in Sensors and Actuators B: Chemical
- Vol. 441
An improved electrochemical biosensing device architecture with ultra-thin and surface-modified dielectric stacking structure
- 18 May 2025
- Published by Elsevier in Sensors and Actuators B: Chemical
- Vol. 441
Open AccessAn improved electrochemical biosensing device architecture with ultra-thin and surface-modified dielectric stacking structure
- Open Access
- 28 October 2024
- Published by MDPI AG in Preprints.org
Food Classification for Dietary Support Using Fine-Grained Visual Recognition with the HERBS Network
Background: Food properties can directly influence individual dietary intake. Therefore, computer vision-based food recognition could be used to estimate meal contents for patients with metabolic diseases. Food recognition based on deep learning can create opportunities for breakthroughs in dietary interventions for personal health management and have rapidly emerged as dietary assessment strategies. Methods: This study proposed a methodology for automatic food recognition based on fine-grained visual classification, the High-temperaturE Refinement and Background Suppression network (HERBS). A technical investigation was conducted involved in the HERBS model on CNFOOD241 benchmark to verify the model effectiveness. The visualization analysis of HERBS was compared with the VGG16 and RepViT on CNFOOD241. Results: The system achieved classification accuracy of 82.72 % and 97.19 % in Top-1 accuracy and Top-5 accuracy respectively. Data showed that HERBS structure widened the attention area and outperformed VGG16 and RepViT on feature maps. Conclusions: Our findings elucidated that the proposed methodology using HERBS model establishes a new benchmark for SOTA performance in food recognition on CNFOOD 241 dataset. This study proved the feasibility of the proposed approach in a challenging food dataset. We further described a vision-language model structure consisting of a multi-modal vision model, language encoder and a multilayer perceptron classifier for food composition recognition.
- 28 October 2024
- Published by MDPI AG in Preprints.org
Open AccessFood Classification for Dietary Support Using Fine-Grained Visual Recognition with the HERBS Network
Background: Food properties can directly influence individual dietary intake. Therefore, computer vision-based food recognition could be used to estimate meal contents for patients with metabolic diseases. Food recognition based on deep learning can create opportunities for breakthroughs in dietary interventions for personal health management and have rapidly emerged as dietary assessment strategies. Methods: This study proposed a methodology for automatic food recognition based on fine-grained visual classification, the High-temperaturE Refinement and Background Suppression network (HERBS). A technical investigation was conducted involved in the HERBS model on CNFOOD241 benchmark to verify the model effectiveness. The visualization analysis of HERBS was compared with the VGG16 and RepViT on CNFOOD241. Results: The system achieved classification accuracy of 82.72 % and 97.19 % in Top-1 accuracy and Top-5 accuracy respectively. Data showed that HERBS structure widened the attention area and outperformed VGG16 and RepViT on feature maps. Conclusions: Our findings elucidated that the proposed methodology using HERBS model establishes a new benchmark for SOTA performance in food recognition on CNFOOD 241 dataset. This study proved the feasibility of the proposed approach in a challenging food dataset. We further described a vision-language model structure consisting of a multi-modal vision model, language encoder and a multilayer perceptron classifier for food composition recognition.