NIRS feature extraction based on deep auto-encoder neural network

Liu, Ting; Zhong-ren, LI; Yu, Chunxia; Qin, Yuhua

doi:10.1016/j.infrared.2017.07.015

Cited by 50 publications

(18 citation statements)

References 11 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Further, many variants were proposed [156]. For example, Liu provided us with a reference example of a deep auto-encoder to extract features for classification tasks [157]. To optimize the performance of feature extraction through obtaining the required network parameters based on proper learning rate, Song proposed a variable learning speed DAE (VLSAE) that adaptively adjusts its learning rate with the help of multi-scale reconstruction error (MRE) and weight update correlation (WUC) [158].…”

Section: Deep Auto Encodermentioning

confidence: 99%

Feature dimensionality reduction: a review

Jia

Sun

Lian

et al. 2022

Complex Intell. Syst.

344

View full text Add to dashboard Cite

As basic research, it has also received increasing attention from people that the “curse of dimensionality” will lead to increase the cost of data storage and computing; it also influences the efficiency and accuracy of dealing with problems. Feature dimensionality reduction as a key link in the process of pattern recognition has become one hot and difficulty spot in the field of pattern recognition, machine learning and data mining. It is one of the most challenging research fields, which has been favored by most of the scholars’ attention. How to implement “low loss” in the process of feature dimension reduction, keep the nature of the original data, find out the best mapping and get the optimal low dimensional data are the keys aims of the research. In this paper, two-dimensionality reduction methods, feature selection and feature extraction, are introduced; the current mainstream dimensionality reduction algorithms are analyzed, including the method for small sample and method based on deep learning. For each algorithm, examples of their application are given and the advantages and disadvantages of these methods are evaluated.

show abstract

Section: Deep Auto Encodermentioning

confidence: 99%

Feature dimensionality reduction: a review

Jia

Sun

Lian

et al. 2022

Complex Intell. Syst.

344

View full text Add to dashboard Cite

show abstract

“…This results in an so called autoencoder architecture. Unsupervised learning without labels has been used for training of autoencoders, which is a major advantage since labeled spectral data can be very expensive to produce [ 28 ]. More precisely, autoencoder architectures are used to reconstruct the input, of which the size is compressed after transforming it to internal hidden layers.…”

Section: Nir Data Processing Using the Neural Network Anninetmentioning

confidence: 99%

SmartSpectrometer—Embedded Optical Spectroscopy for Applications in Agriculture and Industry

Krause

Grüger

Gebauer

et al. 2021

Sensors

View full text Add to dashboard Cite

The ongoing digitization of industry and agriculture can benefit significantly from optical spectroscopy. In many cases, optical spectroscopy enables the estimation of properties such as substance concentrations and compositions. Spectral data can be acquired and evaluated in real time, and the results can be integrated directly into process and automation units, saving resources and costs. Multivariate data analysis is needed to integrate optical spectrometers as sensors. Therefore, a spectrometer with integrated artificial intelligence (AI) called SmartSpectrometer and its interface is presented. The advantages of the SmartSpectrometer are exemplified by its integration into a harvesting vehicle, where quality is determined by predicting sugar and acid in grapes in the field.

show abstract

“…The detection of QS complexes is out of the scope of this paper. The following correct ECG recordings from the MIT-BIH database were used for training and testing: 100,101,102,103,104,105,106,109,112,113,115,116,118,119,121,122,123,201,202,208,209,212,213,214,215,217,219,220,221,222,228,230, 231, 232, and 234. Furthermore, the following correct ECG recordings from the European ST-T database were used for training and testing: e0103, e0104, e0111, e0112, e0113, e0115, e0116, e0118, e0123, e0127, e0136, e0147, e0151, e0154, e0159, e0161, e0166, e0170, e0203, e0204, e0206, e0207, e0208, e0210, e0212, e0303, e0306, e0404, e0406, e0408, e0409, e0410, e0411, e0417, e0418, e0509, e0601, e0606, e0607, e0609, e0610, e0611, e0612, e0613, e0615, e0704, e0818, and e1304.…”

Section: Data Preparationmentioning

confidence: 99%

“…For the MIT-BIH NST, the following recordings are used for training and testing: 100,104,108,113,117,122,201,207,212,217,222,231,101,105,109,114,118,123,208,213,219,223,232,102,106,111,115,119,124,203,209,214,220,228,233,103,107,112,116,121,200,205,210,215,221,230, and 234. The QRS complex inverter is applied to the MIT-BIH as it has many inverted QRS complexes.…”

Section: Mit-bih Nstmentioning

confidence: 99%

See 1 more Smart Citation

Applications of Machine Learning

Rudström¹

2020

Algorithms for Intelligent Systems

View full text Add to dashboard Cite

In this thesis, many machine learning algorithms were applied to electrocardiogram (ECG), spectral analysis, and Field Programmable Gate Arrays (FPGAs). In ECG, QRS complexes are useful for measuring the heart rate and for the segmentation of ECG signals. QRS complexes were detected using WaveletCNN Autoencoder filters and ConvLSTM detectors. The WaveletCNN Autoencoders filters the ECG signals using the wavelet filters, while the ConvLSTM detects the spatial temporal patterns of the QRS complexes. For the spectral analysis topic, the detection of chemical compounds using spectral analysis is useful for identifying unknown substances. However, spectral analysis algorithms require vast amounts of data. To solve this problem, B-spline neural networks were developed for the generation of infrared and ultraviolet/visible spectras. This allowed for the generation of large training datasets from a few experimental measurements. Graphical Processing Units (GPUs) are good for training and testing neural networks. However, using multiple GPUs together is hard because PCIe bus is not suited for scattering operations and reduce operations. FPGAs are more flexible as they can be arranged in a mesh or toroid or hypercube configuration on the PCB. These configurations provide higher data throughput and results in faster computations. A general neural network framework was written in VHDL for Xilinx FPGAs. It allows for any neural network to be trained or tested on FPGAs.

show abstract

NIRS feature extraction based on deep auto-encoder neural network

Cited by 50 publications

References 11 publications

Feature dimensionality reduction: a review

Feature dimensionality reduction: a review

SmartSpectrometer—Embedded Optical Spectroscopy for Applications in Agriculture and Industry

Applications of Machine Learning

Contact Info

Product

Resources

About