scholarly journals A Novel Two-Stage Refine Filtering Method for EEG-Based Motor Imagery Classification

2021 ◽  
Vol 15 ◽  
Author(s):  
Yuxin Yan ◽  
Haifeng Zhou ◽  
Lixin Huang ◽  
Xiao Cheng ◽  
Shaolong Kuang
Keyword(s):  
Motor Imagery ◽  
Training Data ◽  
Cerebral Stroke ◽  
Common Disease ◽  
Two Stage ◽  
Filtering Method ◽  
Feature Extracting ◽  
Kappa Value ◽  
Promising Application ◽  
Filtering Effect

Cerebral stroke is a common disease across the world, and it is a promising method to recognize the intention of stroke patients with the help of brain–computer interface (BCI). In the field of motor imagery (MI) classification, appropriate filtering is vital for feature extracting of electroencephalogram (EEG) signals and consequently influences the accuracy of MI classification. In this case, a novel two-stage refine filtering method was proposed, inspired by Gradient-weighted Class Activation Mapping (Grad-CAM), which uses the gradients of any target concept flowing into the final convolutional layer to highlight the important part of training data for predicting the concept. In the first stage, MI classification was carried out and then the frequency band to be filtered was calculated according to the Grad-CAM of the MI classification results. In the second stage, EEG was filtered and classified for a higher classification accuracy. To evaluate the filtering effect, this method was applied to the multi-branch neural network proposed in our previous work. Experiment results revealed that the proposed method reached state-of-the-art classification kappa value levels and acquired at least 3% higher kappa values than other methods This study also proposed some promising application scenarios with this filtering method.

scholarly journals Patch-Based Change Detection Method for SAR Images with Label Updating Strategy

2021 ◽  
Vol 13 (7) ◽  
pp. 1236
Author(s):  
Yuanjun Shu ◽  
Wei Li ◽  
Menglong Yang ◽  
Peng Cheng ◽  
Songchen Han
Keyword(s):  
Neural Networks ◽  
Deep Learning ◽  
Change Detection ◽  
Detection Method ◽  
Training Data ◽  
Synthetic Aperture ◽  
Learning Method ◽  
Sar Images ◽  
Two Stage ◽  
Second Stage

Convolutional neural networks (CNNs) have been widely used in change detection of synthetic aperture radar (SAR) images and have been proven to have better precision than traditional methods. A two-stage patch-based deep learning method with a label updating strategy is proposed in this paper. The initial label and mask are generated at the pre-classification stage. Then a two-stage updating strategy is applied to gradually recover changed areas. At the first stage, diversity of training data is gradually restored. The output of the designed CNN network is further processed to generate a new label and a new mask for the following learning iteration. As the diversity of data is ensured after the first stage, pixels within uncertain areas can be easily classified at the second stage. Experiment results on several representative datasets show the effectiveness of our proposed method compared with several existing competitive methods.

Switching filtering method for interactive systems

2019 ◽  
Vol 233 (10) ◽  
pp. 1340-1350
Author(s):  
Junyi Hou ◽  
Lei Yu ◽  
Yifan Fang ◽  
Shumin Fei
Keyword(s):  
Real Time ◽  
Subjective Evaluation ◽  
Objective Evaluation ◽  
Interactive Systems ◽  
Interactive System ◽  
The Real ◽  
Noise Interference ◽  
Filtering Method ◽  
Time Performance ◽  
Filtering Effect

Aiming at the problem that the mixed noise interference caused by the mixed projection noise system is not accurate and the real-time performance is poor, this article proposes an adaptive system switching filtering method based on Bayesian estimation switching rules. The method chooses joint bilateral filtering and improved adaptive median filtering as the filtering subsystems and selects the sub-filtering system suitable for the noise by switching rules to achieve the purpose of effectively removing noise. The simulation experiment was carried out by the self-developed human–computer interactive projection image system platform. Through the subjective evaluation, objective evaluation, and running time comparison analysis, a better filtering effect was achieved, and the balance between the filtering precision and the real-time performance of the interactive system was well obtained. Therefore, the proposed method can be widely applied to various human–computer interactive image filtering systems.

scholarly journals A Speckle Filtering Method Based on Hypothesis Testing for Time-Series SAR Images

2018 ◽  
Vol 10 (9) ◽  
pp. 1383 ◽  
Author(s):  
Jili Yuan ◽  
Xiaolei Lv ◽  
Rui Li
Keyword(s):  
Time Series ◽  
Hypothesis Testing ◽  
Speckle Noise ◽  
Sar Images ◽  
Suppression Effect ◽  
Filtering Method ◽  
Information Preservation ◽  
Spatiotemporal Information ◽  
Test Algorithm ◽  
Filtering Effect

To improve the suppression effect for the speckle noise of synthetic aperture radar (SAR) images and the ability of spatiotemporal information preservation of the filtered image without losing the spatial resolution, a novel multitemporal filtering method based on hypothesis testing is proposed in this paper. A framework of a two-step similarity measure strategy is adopted to further enhance the filtering results. Firstly, bi-date analysis using a two-sample Kolmogorov-Smirnov (KS) test is conducted in step 1 to extract homogeneous patches for 3-D patch stacks generation. Subsequently, the similarity between patch stacks is compared by a sliding time-series likelihood ratio (STSLR) test algorithm in step 2, which utilizes the multi-dimensional data structure of the stacks to improve the accuracy of unchanged pixels detection. Finally, the filtered values are obtained by averaging the similar pixels in time-series. The experimental results and analysis of two multitemporal datasets acquired by TerraSAR-X show that the proposed method outperforms the other typical methods with regard to the overall filtering effect, especially in terms of the consistency between the filtered images and the original ones. Furthermore, the performance of the proposed method is also discussed by analyzing the results from step 1 and step 2.

scholarly journals Intelligent Classification Method for Tunnel Lining Cracks Based on PFC-BP Neural Network

2020 ◽  
Vol 2020 ◽  
pp. 1-12
Author(s):  
Hao Ding ◽  
Xinghong Jiang ◽  
Ke Li ◽  
Hongyan Guo ◽  
Wenfeng Li
Keyword(s):  
Neural Network ◽  
Bp Neural Network ◽  
Structural Parameters ◽  
Crack Depth ◽  
Tunnel Lining ◽  
Parameter Analysis ◽  
Training Data ◽  
Concrete Lining ◽  
Common Disease ◽  
Lining Structure

Tunnel lining crack is the most common disease and also the manifestation of other diseases, which widely exists in plain concrete lining structure. Proper evaluation and classification of engineering conditions directly relate to operation safety. Particle flow code (PFC) calculation software is applied in this study, and the simulation reliability is verified by using the laboratory axial compression test and 1 : 10 model experiment to calibrate the calculation parameters. Parameter analysis is carried out focusing on the load parameters, structural parameters, dimension, and direction which affect the crack diseases. Based on that, an evaluation index system represented by tunnel buried depth (H), crack position (P), crack length (L), crack width (W), crack depth (D), and crack direction (A) is put forward. The training data of the back propagation (BP) neural network which takes load-bearing safety and crack stability as the evaluation criteria are obtained. An expert system is introduced into the BP neural network for correction of prediction results, realizing classified dynamic optimization of complex engineering conditions. The results of this study can be used to judge the safety state of cracked lining structure and provide guidance to the prevention and control of crack diseases, which is significant to ensure the safety of tunnel operation.

scholarly journals Multisubject Learning for Common Spatial Patterns in Motor-Imagery BCI

2011 ◽  
Vol 2011 ◽  
pp. 1-9 ◽  
Author(s):  
Dieter Devlaminck ◽  
Bart Wyns ◽  
Moritz Grosse-Wentrup ◽  
Georges Otte ◽  
Patrick Santens
Keyword(s):  
Motor Imagery ◽  
Spatial Filter ◽  
Training Data ◽  
Machine Learning Techniques ◽  
Data Sets ◽  
Calibration Data ◽  
Common Spatial Pattern ◽  
Computer Interfaces ◽  
Learning Techniques ◽  
The Common

Motor-imagery-based brain-computer interfaces (BCIs) commonly use the common spatial pattern filter (CSP) as preprocessing step before feature extraction and classification. The CSP method is a supervised algorithm and therefore needs subject-specific training data for calibration, which is very time consuming to collect. In order to reduce the amount of calibration data that is needed for a new subject, one can apply multitask (from now on called multisubject) machine learning techniques to the preprocessing phase. Here, the goal of multisubject learning is to learn a spatial filter for a new subject based on its own data and that of other subjects. This paper outlines the details of the multitask CSP algorithm and shows results on two data sets. In certain subjects a clear improvement can be seen, especially when the number of training trials is relatively low.

Classification of Peer-to-Peer Traffic Using a Two-Stage Window-Based Classifier with Fast Decision Tree and IP Layer Attributes

2011 ◽  
pp. 174-188
Author(s):  
Bijan Raahemi ◽  
Ali Mumtaz
Keyword(s):  
Decision Tree ◽  
High Accuracy ◽  
Peer To Peer ◽  
Training Data ◽  
Two Stage ◽  
Layer 4 ◽  
Using Data ◽  
P2p Applications ◽  
Fast Decision

This paper presents a new approach using data mining techniques, and in particular a two-stage architecture, for classification of Peer-to-Peer (P2P) traffic in IP networks where in the first stage the traffic is filtered using standard port numbers and layer 4 port matching to label well-known P2P and NonP2P traffic. The labeled traffic produced in the first stage is used to train a Fast Decision Tree (FDT) classifier with high accuracy. The Unknown traffic is then applied to the FDT model which classifies the traffic into P2P and NonP2P with high accuracy. The two-stage architecture not only classifies well-known P2P applications, but also classifies applications that use random or non-standard port numbers and cannot be classified otherwise. The authors captured the internet traffic at a gateway router, performed pre-processing on the data, selected the most significant attributes, and prepared a training data set to which the new algorithm was applied. Finally, the authors built several models using a combination of various attribute sets for different ratios of P2P to NonP2P traffic in the training data.

An Approach to Tone Recognition of Mandarin Speech Based-On Two-Stage Model

2011 ◽  
Vol 145 ◽  
pp. 297-303
Author(s):  
Wei Chih Hsu ◽  
Jung Nan Sun ◽  
Huai I Wang
Keyword(s):  
Standard Deviation ◽  
Recognition Rate ◽  
Threshold Value ◽  
Recognition Algorithm ◽  
Training Data ◽  
Pitch Contour ◽  
Two Stage ◽  
Second Stage ◽  
Fundamental Frequencies ◽  
Tone Recognition

After inspecting the pitch contours of tone 1 of Mandarin speech, we found that the pitch contour of tone 1 consists of upward and downward line segments, while it is supposed that the contour of tone 1 is flat. Our study also found that tone 1 tends to be recognized as other three tones if the recognition algorithm used is based on the tone contour slope or shape. According to our experiments, we conclude that the recognition rate of the tones would be improved if two stage tone recognition scheme is conducted. At the first stage, tone one is recognized out and then the other three tones are identified at the second stage. The fundamental frequencies of input Mandarin speech of tone 1 are first retrieved from the training data and then a threshold value relating to standard deviation of fundamental frequencies is determined. In the first recognition stage, if the statistic standard deviation of fundamental frequencies is less than the determined threshold, the Mandarin speech is recognized as tone one. The input Mandarin speech which is not classified as tone 1 are the recognition targets of the second recognition stage. In the second stage, a so-called linear gradient analysis is conducted, and the tones are identified according to the derived positive or negative linear gradients. Our proposed recognition method is superior to traditional methods of Mandarin tone recognition in terms of effectiveness and recognition rate. Some experiments to prove the necessity of conducting two recognition stages will be described in detail.

Sign in / Sign up

Export Citation Format

Share Document