A CNN-Inspired Mixed Signal Processor Based on Tunnel Transistors

Sedighi, Behnam; Palit, Indranil; Hu, Xiaobo Sharon; Nahas, Joseph; Niemier, Michael

doi:10.7873/date.2015.0053

“…It is a deep neural network method that simulates the deep hierarchal structure of human vision and has been successfully applied in image classification, natural language processing (NLP) and speech recognition (Palaz & Collobert, 2015; Sharif Razavian et al, 2014; Yin et al, 2017). Due to its proficiency in automatic feature extraction, CNN is also used to design advanced signal analysis methods (Kwon, Shin & Kim, 2018; Sedighi et al, 2015). For example, Kiranyaz, Ince & Gabbouj (2015) used CNN for ECG classification.…”

Section: Methodsmentioning

confidence: 99%

“…Compared to traditional machine learning methods, a CNN does not require hand-crafted features, and can automatically extract effective features through hierarchical layers. It has been successfully applied in speech recognition (Abdel-Hamid et al, 2012; Palaz & Collobert, 2015), image classification (Sharif Razavian et al, 2014; Wei et al, 2016), signal analysis (Kwon, Shin & Kim, 2018; Sedighi et al, 2015) and other fields. LeNet-5 is one CNN implementation with relatively few parameters and good performance (El-Sawy, Hazem & Loey, 2016; LeCun, 2015; Wen et al, 2018).…”

Section: Introductionmentioning

confidence: 99%

Sleep apnea detection from a single-lead ECG signal with automatic feature-extraction through a modified LeNet-5 convolutional neural network

Wang

¹

,

Lu

²

,

Shen

³

et al. 2019

PeerJ

View full text Add to dashboard Cite

Sleep apnea (SA) is the most common respiratory sleep disorder, leading to some serious neurological and cardiovascular diseases if left untreated. The diagnosis of SA is traditionally made using Polysomnography (PSG). However, this method requires many electrodes and wires, as well as an expert to monitor the test. Several researchers have proposed instead using a single channel signal for SA diagnosis. Among these options, the ECG signal is one of the most physiologically relevant signals of SA occurrence, and one that can be easily recorded using a wearable device. However, existing ECG signal-based methods mainly use features (i.e. frequency domain, time domain, and other nonlinear features) acquired from ECG and its derived signals in order to construct the model. This requires researchers to have rich experience in ECG, which is not common. A convolutional neural network (CNN) is a kind of deep neural network that can automatically learn effective feature representation from training data and has been successfully applied in many fields. Meanwhile, most studies have not considered the impact of adjacent segments on SA detection. Therefore, in this study, we propose a modified LeNet-5 convolutional neural network with adjacent segments for SA detection. Our experimental results show that our proposed method is useful for SA detection, and achieves better or comparable results when compared with traditional machine learning methods.

show abstract

“…It is a deep neural network method that simulates the deep hierarchal structure of human vision and has been successfully applied in image classification, natural language processing (NLP) and speech recognition (Palaz & Collobert 2015;Sharif Razavian et al 2014;Yin et al 2017). Due to its proficiency in automatic feature extraction, CNN is also used to design advanced signal analysis methods (Kwon et al 2018;Sedighi et al 2015). For example, (Kiranyaz et al 2015) used CNN for ECG classification.…”

Section: Convolutional Neural Networkmentioning

confidence: 99%

“…Compared to traditional machine learning methods, a CNN does not require hand-crafted features, and can automatically extract effective features through hierarchical layers. It has been successfully applied in speech recognition (Abdel-Hamid et al 2012;Palaz & Collobert 2015), image classification (Sharif Razavian et al 2014;Wei et al 2016), signal analysis (Kwon et al 2018;Sedighi et al 2015) and other fields. LeNet-5 is one CNN implementation with relatively few parameters and good performance (El-Sawy et al 2016;LeCun 2015;Wen et al 2018).…”

Section: Introductionmentioning

confidence: 99%

Peer Review #2 of "Sleep apnea detection from a single-lead ECG signal with automatic feature-extraction through a modified LeNet-5 convolutional neural network (v0.1)"

Vaquerizo-Villar¹

2019

0

View full text Add to dashboard Cite

Sleep apnea (SA) is the most common respiratory sleep disorder, leading to some serious neurological and cardiovascular diseases if left untreated. The diagnosis of SA is traditionally made using Polysomnography (PSG). However, this method requires many electrodes and wires, as well as an expert to monitor the test. Several researchers have proposed instead using a single channel signal for SA diagnosis. Among these options, the ECG signal is one of the most physiologically relevant signals of SA occurrence, and one that can be easily recorded using a wearable device. However, existing ECG signal-based methods mainly use features (i.e. frequency domain, time domain, and other nonlinear features) acquired from ECG and its derived signals in order to construct the model. This requires researchers to have rich experience in ECG, which is not common. A convolutional neural network (CNN) is a kind of deep neural network that can automatically learn effective feature representation from training data and has been successfully applied in many fields. Meanwhile, most studies have not considered the impact of adjacent segments on SA detection. Therefore, in this study, we propose a modified LeNet-5 convolutional neural network with adjacent segments for SA detection. Our experimental results show that our proposed method is useful for SA detection, and achieves better or comparable results when compared with traditional machine learning methods.

show abstract

“…Even though, a lot of MOS improvement techniques such as Junctionless MOS on SOI and other structural modifications has been proposed [7][8][9] but still it is insufficient to withstand the power constraints in resource limited applications. Alternatively, tunnel field effect transistor (TFET) [10] is emerged as a potential descendant for ultra-low power applications because of its better subthreshold swing (SS), small OFF state leakage current and low voltage operation. Among all TFETs, nano-wire TFET offers immunity against SCEs, better electrostatic control due to warping of the gate around the channel, no corner effects and carrier confinement near oxide interface as compared to the bulk MOSFETs.…”

Section: Introductionmentioning

confidence: 99%

Implementation of ∑Δ ADC using electrically doped III‐V ternary alloy semiconductor nano‐wire TFET

Rajan

¹

,

Patel

²

,

Sharma

³

et al. 2020

Micro & Nano Letters

11

0

View full text Add to dashboard Cite

In this work, a fast and low‐power sigma delta )(∑normalΔ analogue‐to‐digital converter (ADC) has been developed using a hetero‐material electrically‐doped nano‐wire tunnel field effect transistor (HM‐ED‐NW‐TFET) for the first time. The better gate controllability of nano‐wire and immunity against process variations of electrically doped tunnel field effect transistor (TFET) enhances resolution. In this regard, the first step that has been performed is the material engineering using normalAlxnormalGa1−xnormalSb/normalGaAs1−yPy, to achieve significant driving current at low subthreshold swing and high InormalON/InormalOFF ratio. Secondly, the mole fraction is optimised to upgrade the critical analogue component – the op‐amp. Also, drain under lapping is included in p‐TFET to bring its characteristics as close to n‐TFET. Latter, the look‐up tables of the proposed device has been generated which is used to develop individual block of ∑normalΔ ADC in Cadence. The blocks are well verified and integrated into final ∑normalΔ ADC and its performance is evaluated. Hence, this work has explored the inherent merits of HM‐ED‐NW‐TFET in the development of low power and fast ∑normalΔ ADC by virtue of a high slew rate op‐amp. Therefore, this work contributes a novel approach to explore the characteristics of emerging devices in mixed signal applications.

show abstract

A CNN-Inspired Mixed Signal Processor Based on Tunnel Transistors

Cited by 4 publications

References 0 publications

Sleep apnea detection from a single-lead ECG signal with automatic feature-extraction through a modified LeNet-5 convolutional neural network

Sleep apnea detection from a single-lead ECG signal with automatic feature-extraction through a modified LeNet-5 convolutional neural network

Peer Review #2 of "Sleep apnea detection from a single-lead ECG signal with automatic feature-extraction through a modified LeNet-5 convolutional neural network (v0.1)"

Implementation of ∑Δ ADC using electrically doped III‐V ternary alloy semiconductor nano‐wire TFET

Contact Info

Product

Resources

About