Feature extraction for deep neural networks based on decision boundaries

Woo, Seongyoun; Lee, Chulhee

doi:10.1117/12.2263172

Cited by 4 publications

(3 citation statements)

References 11 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The traditional method to solve the problem of missing training samples is to expand the existing labeled samples by rotating, translating, mirror transformation, and adding Gauss noise to the original labeled samples, and then generate new labeled samples to increase the number of samples. However, from the information point of view, although the diversity of samples is increased after data expansion, the amount of essential information beneficial to the training network has not increased [18,19] (Figs. 2 and 3).…”

Section: Using Transfer Learning To Solve the Problem Of Missing Traimentioning

confidence: 99%

A security risk plan search assistant decision algorithm using deep neural network combined with two-stage similarity calculation

Fang

et al. 2019

Pers Ubiquit Comput

View full text Add to dashboard Cite

In view of the nonlinearity and uncertainty of safety accident risk assessment, firstly, based on the deep neural network, the training criterion of the network is changed, and the triplet convolutional neural network with the similarity measure as the cost function is proposed. The inactive multi-scale set features are extracted from them, so that the semantic features obtained by learning are suitable for security risk image retrieval. In the image retrieval application, the training samples of the retrieved data set are not enough to train a large network, and the innovative application of migration learning to security risk image retrieval proposes to train the network with data sets similar to the retrieved data sets. Then based on the traditional nearest neighbor algorithm, this paper proposes a case similarity calculation method based on two-dimensional structure of structural similarity and attribute similarity, input the characteristic attribute value of the current emergency event, and conduct similar case retrieval. The final calculation returns the historical case and its solution that the user is most similar to the currently entered incident feature. The experiment proves that the maximum relative error between the output of the network and the expected output value is 5.17%, and the minimum relative error is 1.38%, which has high accuracy.

show abstract

Section: Using Transfer Learning To Solve the Problem Of Missing Traimentioning

confidence: 99%

A security risk plan search assistant decision algorithm using deep neural network combined with two-stage similarity calculation

Fang

et al. 2019

Pers Ubiquit Comput

View full text Add to dashboard Cite

show abstract

“…Feature extraction is one of the fundamental prerequisite processes used in machinelearning classification and pattern recognition applications. As such, there has been extensive research effort into the development of an accurate feature extraction technique [1][2][3]. Feature extraction is described as a process through which a useful set of attributes is derived from the raw time-series data while removing its redundant information.…”

Section: Introductionmentioning

confidence: 99%

Feature Extraction of Time-Series Data Using DWT and FFT for Ballscrew Condition Monitoring

Alegeh

Thottoli

Mian

et al. 2021

Advances in Transdisciplinary Engineering

View full text Add to dashboard Cite

This paper investigates the use of the discrete wavelet transform (DWT) and Fast Fourier Transform (FFT) to improve the quality of extracted features for machine learning. The case study in this paper is detecting the health state of the ballscrew of a gantry type machine tool. For the implementation of the algorithm for feature extraction, wavelet is first applied to the data, followed by FFT and then useful features are extracted from the resultant signal. The extracted features were then used in various machine learning algorithms like decision tree, K-nearest neighbour (KNN) and support vector machine (SVM) for binary classification of the ballscrew state. The result shows significant improvement in the classification accuracy after the wavelet transform and FFT has been performed on the data.

show abstract

“…With the surging of deep learning, neural networks can model multiple events and learn richer representations that have the potential to learn better models of nonlinear data [15][16][17]. With multiple layers, deep neural networks (DNNs) [18,19] perform well on decision boundary and feature engineering problems by using a massive amount of data [20]. In recent years, a deep neural network hidden Markov model (DNN-HMM) has been proposed as a novel hybrid architecture and has been widely used on acoustic learning [21][22][23].…”

Section: Introductionmentioning

confidence: 99%

A Hybrid Hidden Markov Model for Pipeline Leakage Detection

Zhang

Chen

2021

Applied Sciences

View full text Add to dashboard Cite

In this paper, a deep neural network hidden Markov model (DNN-HMM) is proposed to detect pipeline leakage location. A long pipeline is divided into several sections and the leakage occurs in different section that is defined as different state of hidden Markov model (HMM). The hybrid HMM, i.e., DNN-HMM, consists of a deep neural network (DNN) with multiple layers to exploit the non-linear data. The DNN is initialized by using a deep belief network (DBN). The DBN is a pre-trained model built by stacking top-down restricted Boltzmann machines (RBM) that compute the emission probabilities for the HMM instead of Gaussian mixture model (GMM). Two comparative studies based on different numbers of states using Gaussian mixture model-hidden Markov model (GMM-HMM) and DNN-HMM are performed. The accuracy of the testing performance between detected state sequence and actual state sequence is measured by micro F1 score. The micro F1 score approaches 0.94 for GMM-HMM method and it is close to 0.95 for DNN-HMM method when the pipeline is divided into three sections. In the experiment that divides the pipeline as five sections, the micro F1 score for GMM-HMM is 0.69, while it approaches 0.96 with DNN-HMM method. The results demonstrate that the DNN-HMM can learn a better model of non-linear data and achieve better performance compared to GMM-HMM method.

show abstract

Feature extraction for deep neural networks based on decision boundaries

Cited by 4 publications

References 11 publications

A security risk plan search assistant decision algorithm using deep neural network combined with two-stage similarity calculation

A security risk plan search assistant decision algorithm using deep neural network combined with two-stage similarity calculation

Feature Extraction of Time-Series Data Using DWT and FFT for Ballscrew Condition Monitoring

A Hybrid Hidden Markov Model for Pipeline Leakage Detection

Contact Info

Product

Resources

About