Regression with Deep Learning for Sensor Performance Optimization

Vaila, Ruthvik; Lloyd, Denver; Tetz, Kevin

doi:10.1109/wmed49473.2021.9425222

Cited by 3 publications

(4 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Tempotron rule can effectively utilize supervised signals to train the weights so that the output neuron can fire according to its class label. Zhao et al 26 and Xiao et al 31 44 used SNN with R-STDP learning rule to process the features and they obtained 90.71% classification accuracy on N-MNIST dataset, which is 10.08% higher than the performance of Iyer's method. 43 The SNN with R-STDP learning rule improves the performance of SNN with STDP learning rule on event stream datasets.…”

Section: Event-based Feature Recognition Methods For Event Cameramentioning

confidence: 99%

“…It helps the network to adjust weights by combining local STDP learning rule with a global reward signal and enables it to detect rare and diagnostic features. Vaila et al 44 . used SNN with R-STDP learning rule to process the features and they obtained 90.71% classification accuracy on N-MNIST dataset, which is 10.08% higher than the performance of Iyer’s method 43 .…”

Section: Related Workmentioning

confidence: 97%

See 1 more Smart Citation

Event camera object recognition using spatiotemporal event time surface and reward-modulated spike-timing-dependent plasticity learning rule

Zhou,

Zhang

2024

J. Electron. Imag.

View full text Add to dashboard Cite

Event cameras record moving objects with asynchronous event streams. It remains a challenge to make full use of the spatiotemporal information of event streams to extract high-quality features and to make event camera object recognition. We propose an event-based event camera object recognition system, which includes a denoising module and an object recognition module. The denoising module removes noise events using spatiotemporal information of the events. The object recognition module extracts primary spatiotemporal features based on the event time surface prototypes obtained by clustering, then further extracts complex diagnostic features, and makes object recognition using spiking neural networks with reward-modulated spike-timing-dependent plasticity learning rule. Experimental results demonstrate that our system has better performance than baseline methods on five popular event camera datasets, especially on datasets with rich spatiotemporal dynamics. Our denoising method improves the noise robustness of our event camera object recognition system on high-speed moving datasets that are greatly affected by noises. Moreover, our method has much better recognition ability than baseline methods when using short input event streams. Our method is very beneficial for developing event-based event camera object recognition algorithm when event streams are short or noises are serious.

show abstract

Section: Event-based Feature Recognition Methods For Event Cameramentioning

confidence: 99%

Section: Related Workmentioning

confidence: 97%

Event camera object recognition using spatiotemporal event time surface and reward-modulated spike-timing-dependent plasticity learning rule

Zhou,

Zhang

2024

J. Electron. Imag.

View full text Add to dashboard Cite

show abstract

“…And we find that the fitted constant is uncertain and it changes with the wavelength and the fraction of Al by calculating the refractive index with the equation above. So, to get a reliable parameter, we construct a set of algorithms according to the deep learning method [4] to get a reliable refrac--tive index based on the wavelength and Al's fraction. The training set and test set that are used in this the deep learning algorithm are obtained from the network.…”

Section: Refractive Indexmentioning

confidence: 99%

Study on small spacing total internal reflection quantum well infrared focal plane array device

shao

Li²,

Hong³

et al. 2023

Earth and Space: From Infrared to Terahertz (ESIT 2022)

View full text Add to dashboard Cite

Since infrared detection was proposed, infrared detectors have been to the 3rd generation . Both of the small size and large scale are the requirements for the development of infrared detectors. At present, the focal plane array devices on the market have a relatively high difference between the response wavelength and the size of pixel and therefore, when dealing with long wavelength detection, there will be a large size FPA. The quantum well infrared focal plane array devices fabricated by GaAs/AlGaAs have been widely used due to high uniformity and mature technology, but in the field of long-wave and very-long-wave detection, if the traditional grating coupling structure is used, because the wavelength and the size of the pixel are close, a strong diffraction effect will occur before the infrared radiation reaches the active region. This will lead to significant crosstalk and errors. Therefore, using grating diffraction as the quantum well coupling mechanism limits the size of the pixel and as a coupling mechanism, it is difficult to achieve due to small pixel size. Therefore, this paper adopts the total internal reflection type coupling structure proposed by K.K.Choi, and compare it with the traditional grating coupling structure to study how this structure improves the performance when detecting long-wavelength and reducing the size of pixels. In this paper, a quantum well infrared focal plane array with a pixel size of 640×512, a center-to-center distance of 15 μm and a response wavelength of 10.55 μm is fabricated by using GaAs/AlGaAs and melting photoresist technology which is different with the method proposed by K.K.Choi to fabricate this structure. The FDTD-based open source field simulation software MEEP is used to simulate the field distribution of the devices and evaluate its performance about fighting against optical crosstalk of this structure and compare its performance with the grating coupling structure. In this paper, we also use MEEP to explore the influence of the reflection angle, the position of the active region, and the period of the quantum well on the distribution of the field inside the device, and calculate the electromagnetic wave’s energy of the active region as the evaluation factor. These results shows that these geometrical factors restrict each other, therefore, to produce a good QWIP FPA based on the total internal reflective structure, one need to take these factors into account to tune these parameters to maximize the optimized performance. The main content of the research is to fabricate a 640×512 total internal reflection quantum well infrared focal plane array device, and use the MEEP to study the electrical field distribution inside the device with related geometric factors.

show abstract

“…The maximum current value is 0.558 A and the minimum value is 0 A. The higher the current value, the greater the Joule heat, and the NiCr thin film is destroyed when it exceeds the melting point 1400 • C. In this paper, we propose a method to thermally optimize the trimming pattern based on the policy above, taking advantage of machine learning which is also widely used for regression analysis [32][33] [34]. In the following section, we present the process step of the proposed methodology, comparing a conventional way of L-cut trimming.…”

Section: Introductionmentioning

confidence: 99%

A thermally optimizing method of thin film resistor trimming with machine learning

Arima

Hidaka

Watanabe

et al. 2023

IEICE Electron. Express

View full text Add to dashboard Cite

This paper proposes a novel trimming method of thin-film resistors dedicated to the power modules. Thin-film resistors are utilized for applications including snubber circuits which avoids unwanted ringing appearing in the output voltage of power transistors. Our previous works have revealed that the applicable voltage of a NiCr thin-film resistor is thermally limited, and then the indispensable trimming process affects the degree of temperature rise in use. In this paper, we first formulate the relationship between the target resistance value and the trim dimensions using machine learning. With the obtained equation, we propose a new trimming method, which enables the trimmed pattern to reduce the variability of the maximum temperature rise. The experimental results show that the proposed trimming method can suppress the estimated range of the maximum temperature from 619 K to 179 K.

show abstract

Regression with Deep Learning for Sensor Performance Optimization

Cited by 3 publications

References 10 publications

Event camera object recognition using spatiotemporal event time surface and reward-modulated spike-timing-dependent plasticity learning rule

Event camera object recognition using spatiotemporal event time surface and reward-modulated spike-timing-dependent plasticity learning rule

Study on small spacing total internal reflection quantum well infrared focal plane array device

A thermally optimizing method of thin film resistor trimming with machine learning

Contact Info

Product

Resources

About