Deep-Learning-Based Analysis of Electronic Skin Sensing Data DOI Creative Commons

Yu-Chen Guo,

Xidi Sun,

Lulu Li

и другие.

Sensors, Год журнала: 2025, Номер 25(5), С. 1615 - 1615

Опубликована: Март 6, 2025

E-skin is an integrated electronic system that can mimic the perceptual ability of human skin. Traditional analysis methods struggle to handle complex e-skin data, which include time series and multiple patterns, especially when dealing with intricate signals real-time responses. Recently, deep learning techniques, such as convolutional neural network, recurrent transformer methods, provide effective solutions automatically extract data features recognize significantly improving data. Deep not only capable handling multimodal but also response personalized predictions in dynamic environments. Nevertheless, problems insufficient annotation high demand for computational resources still limit application e-skin. Optimizing algorithms, efficiency, exploring hardware-algorithm co-designing will be key future development. This review aims present techniques applied inspiration subsequent researchers. We first summarize sources characteristics models applicable their applications analysis. Additionally, we discuss use e-skin, particularly health monitoring human-machine interactions, explore current challenges development directions.

Язык: Английский

Deep-Learning-Based Analysis of Electronic Skin Sensing Data DOI Creative Commons

Yu-Chen Guo,

Xidi Sun,

Lulu Li

и другие.

Sensors, Год журнала: 2025, Номер 25(5), С. 1615 - 1615

Опубликована: Март 6, 2025

E-skin is an integrated electronic system that can mimic the perceptual ability of human skin. Traditional analysis methods struggle to handle complex e-skin data, which include time series and multiple patterns, especially when dealing with intricate signals real-time responses. Recently, deep learning techniques, such as convolutional neural network, recurrent transformer methods, provide effective solutions automatically extract data features recognize significantly improving data. Deep not only capable handling multimodal but also response personalized predictions in dynamic environments. Nevertheless, problems insufficient annotation high demand for computational resources still limit application e-skin. Optimizing algorithms, efficiency, exploring hardware-algorithm co-designing will be key future development. This review aims present techniques applied inspiration subsequent researchers. We first summarize sources characteristics models applicable their applications analysis. Additionally, we discuss use e-skin, particularly health monitoring human-machine interactions, explore current challenges development directions.

Язык: Английский

Процитировано

0