Auto-Encoder Based for High Spectral Dimensional Data Classification and Visualization

Zhu, Jiasong; Wu, Lei; Hao, Hongxing; Song, Xiaohong; Lu, Yi

doi:10.1109/dsc.2017.32

Cited by 12 publications

(6 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…To learn high-level representation from data, the work [221] proposed a combination of multi-layer AEs with maximum noise fraction which reduces the spectral dimensionality of HSI, while a softmax logistic regression classifier is employed for HSIC. The study reported in [222] combined multimanifold learning framework proposed by [223] with Counteractive Autoencoder [224] for improved unsupervised HSIC.…”

Section: Autoencoders (Ae)mentioning

confidence: 99%

Hyperspectral Image Classification—Traditional to Deep Models: A Survey for Future Prospects

Ahmad

Shabbir

Roy³

et al. 2022

IEEE J. Sel. Top. Appl. Earth Observations Remote Sensing

190

View full text Add to dashboard Cite

Hyperspectral Imaging (HSI) has been extensively utilized in many real-life applications because it benefits from the detailed spectral information contained in each pixel. Notably, the complex characteristics i.e., the nonlinear relation among the captured spectral information and the corresponding object of HSI data make accurate classification challenging for traditional methods. In the last few years, Deep Learning (DL) has been substantiated as a powerful feature extractor that effectively addresses the nonlinear problems that appeared in a number of computer vision tasks. This prompts the deployment of DL for HSI classification (HSIC) which revealed good performance. This survey enlists a systematic overview of DL for HSIC and compared state-of-the-art strategies of the said topic. Primarily, we will encapsulate the main challenges of traditional machine learning for HSIC and then we will acquaint the superiority of DL to address these problems. This survey breakdown the stateof-the-art DL frameworks into spectral-features, spatial-features, and together spatial-spectral features to systematically analyze the achievements (future research directions as well) of these frameworks for HSIC. Moreover, we will consider the fact that

show abstract

Section: Autoencoders (Ae)mentioning

confidence: 99%

Hyperspectral Image Classification—Traditional to Deep Models: A Survey for Future Prospects

Ahmad

Shabbir

Roy³

et al. 2022

IEEE J. Sel. Top. Appl. Earth Observations Remote Sensing

190

View full text Add to dashboard Cite

show abstract

“…Studies show that none of the features individually fulfil the purpose and apparently produced varied data range with wide graphical representations (Table 1 and Figure 2). As it is difficult to visualize and realize a space more than 3 dimensions, the AE has been used to reduce multidimensional space to 2 dimensions without any loss of information [19]. In the latent space, provided by the hidden layers of AE the nonlinearly separable problem has become a linearly separable one.…”

Section: Discussionmentioning

confidence: 99%

“…From input to the hidden layer, having reduced number of nodes, works as encoder i.e., reduces the feature space, followed by increasing number of nodes, working as the decoder. In this work the encoder part of the AE is used to reduce the dimension of the feature space [19,20] without any loss of information regarding classification. The objective of AE is to manipulate the features in the hidden layer in such a way so that X !…”

Section: Reduction Of Feature Space By Auto Encoder (Ae)mentioning

confidence: 99%

Brain Tumour Classification by Machine Learning Applications with Selected Biological Features: Towards A Newer Diagnostic Regime

Ghosh

2020

J. Anal. Oncol.

View full text Add to dashboard Cite

Histopathologically classified low-grade brain tumours show overlapping biological characteristics making them difficult to distinguish. In the present study low-grade brain tumour patient samples of three different histopathological types have been trained through machine learning technique using selected features for its classification. We used specifically the fundamental proliferation, invasion, macrophage infiltration triangle of cancer hallmark with propidium iodide (PI) marked cell-cycle, Ki67 marked proliferative indexing, invasion with MMP2 expression and presence of macrophage/microglia by silver-gold staining, CD11b+ and Iba1+ cell presence as biological parameters. These parameters when trained with proper machine learning protocol through extraction of underling features and represented in a 2D perceivable space are found capable of distinguishing the tumour types. Extracted features from such parameters in a six-dimensional featured space were trained through statistical learning theory while support vector machine (SVM) maximizes their predictive precision. The leave one out (LOO) cross validation process was applied to judge the accuracy of training followed by auto-encoder (AE) to reduce feature dimension at two which is visually perceptible. From the biological features quantified with standard methods it was found impossible to demarcate the three types of low grade brain tumours. However, after training through SVM and LOO cross validation when the sixdimensional featured space had been reduced into two-dimension using AE, the combined output of the features showed clear zonation in that 2D space. This indicates that the overlapping biological characteristics of these tumour types, when trained through proper support vector machine and reduced from multiple to two dimensional space provides a clear patho-clinical classification edge using a combination of common biological features. Hence, machine learning applications may potentially be used as a complementary diagnostic protocol with the conventional practice.

show abstract

“…This ANN was fed by a reduced set of features, tapped from a six-layer DAE. A special form of ANN, DAE [40] can extract effective features from a high-dimensional data in an unsupervised manner. The basic organization of a DAE resembles a funnel shaped structure, with few intermediate layers.…”

Section: E Feature Extraction Using Deep Auto-encoder For Ann Trainingmentioning

confidence: 99%

Improving Photoplethysmographic Measurements Under Motion Artifacts Using Artificial Neural Network for Personal Healthcare

Roy

Gupta

Chandra³

et al. 2018

IEEE Trans. Instrum. Meas.

View full text Add to dashboard Cite

Photoplethysmographic (PPG) measurements are susceptible to motion artifacts (MA) due to movement of the peripheral body parts. In this paper, we present a new approach to identify the MA corrupted PPG beats and then rectify the beat morphology using artificial neural network (ANN). Initially, beat quality assessment was done to identify the clean PPG beats by a pretrained feedback ANN to generate a reference beat template for each person. The PPG data were decomposed using principal component analysis (PCA) and reconstructed using fixed energy retention. A weight coefficient was assigned for each PPG sample in such a way that when they are multiplied, the modified beat morphology matches the reference template. A particle swarm optimization-based technique was utilized to select the best weight vector coefficients to tune another feedback ANN, fed with a set of significant features generated by an auto-encoder from PCA reconstructed data. For real-time implementation, this pretrained ANN was operated in feed-forward mode to directly generate the weight vectors for any subsequent measurements of PPG. The method was validated with PPG data collected from 55 human subjects. An average root-mean-square error of 0.28 and signal-to-noise ratio improvement of 14.54 dB were obtained, with an average improvement of 36% and 47% measurement accuracies on crest time and systolic to diastolic peak height ratio, respectively. With the IEEE Signal Processing Cup 2015 challenge database, Pearson's correlation coefficient between PPG-estimated and ECG-derived heart rates was 0.990. The proposed method can be useful for personal health monitoring applications.Index Terms-Beat quality assessment, deep auto-encoder (DAE), motion artifacts (MA), particle swarm optimization (PSO), photoplethysmography (PPG), signal-to-noise ratio (SNR) improvement.

show abstract

Auto-Encoder Based for High Spectral Dimensional Data Classification and Visualization

Cited by 12 publications

References 13 publications

Hyperspectral Image Classification—Traditional to Deep Models: A Survey for Future Prospects

Hyperspectral Image Classification—Traditional to Deep Models: A Survey for Future Prospects

Brain Tumour Classification by Machine Learning Applications with Selected Biological Features: Towards A Newer Diagnostic Regime

Improving Photoplethysmographic Measurements Under Motion Artifacts Using Artificial Neural Network for Personal Healthcare

Contact Info

Product

Resources

About