scholarly journals A Suppression Method of Concentration Background Noise by Transductive Transfer Learning for a Metal Oxide Semiconductor-Based Electronic Nose

Sensors ◽  
2020 ◽  
Vol 20 (7) ◽  
pp. 1913
Author(s):  
Huixiang Liu ◽  
Qing Li ◽  
Zhiyong Li ◽  
Yu Gu
Keyword(s):  
Transfer Learning ◽  
Environmental Changes ◽  
Oxide Semiconductor ◽  
Target Domain ◽  
Fisher Linear Discriminant ◽  
Linear Discriminant ◽  
Signal Features ◽  
Changes Over Time ◽  
Independence Criterion ◽  
Better Than

Signal drift caused by sensors or environmental changes, which can be regarded as data distribution changes over time, is related to transductive transfer learning, and the data in the target domain is not labeled. We propose a method that learns a subspace with maximum independence of the concentration features (MICF) according to the Hilbert-Schmidt Independence Criterion (HSIC), which reduces the inter-concentration discrepancy of distributions. Then, we use Iterative Fisher Linear Discriminant (IFLD) to extract the signal features by reducing the divergence within classes and increasing the divergence among classes, which helps to prevent inconsistent ratios of different types of samples among the domains. The effectiveness of MICF and IFLD was verified by three sets of experiments using sensors in real world conditions, along with experiments conducted in the authors’ laboratory. The proposed method achieved an accuracy of 76.17%, which was better than any of the existing methods that publish their data on a publicly available dataset (the Gas Sensor Drift Dataset). It was found that the MICF-IFLD was simple and effective, reduced interferences, and deftly managed tasks of transfer classification.

scholarly journals Instance Transfer Learning with Multisource Dynamic TrAdaBoost

2014 ◽  
Vol 2014 ◽  
pp. 1-8 ◽  
Author(s):  
Qian Zhang ◽  
Haigang Li ◽  
Yong Zhang ◽  
Ming Li
Keyword(s):  
Transfer Learning ◽  
Classification Accuracy ◽  
Negative Transfer ◽  
Traditional Learning ◽  
Dynamic Factor ◽  
Similar Distribution ◽  
Target Domain ◽  
Learning Tasks ◽  
Sample Data ◽  
Better Than

Since the transfer learning can employ knowledge in relative domains to help the learning tasks in current target domain, compared with the traditional learning it shows the advantages of reducing the learning cost and improving the learning efficiency. Focused on the situation that sample data from the transfer source domain and the target domain have similar distribution, an instance transfer learning method based on multisource dynamic TrAdaBoost is proposed in this paper. In this method, knowledge from multiple source domains is used well to avoid negative transfer; furthermore, the information that is conducive to target task learning is obtained to train candidate classifiers. The theoretical analysis suggests that the proposed algorithm improves the capability that weight entropy drifts from source to target instances by means of adding the dynamic factor, and the classification effectiveness is better than single source transfer. Finally, experimental results show that the proposed algorithm has higher classification accuracy.

A Novel Face Recognition Algorithm

2013 ◽  
Vol 718-720 ◽  
pp. 2055-2061
Author(s):  
Cai Rang Zhaxi ◽  
Yue Guang Li
Keyword(s):  
Face Recognition ◽  
Recognition Rate ◽  
Feature Space ◽  
Recognition Algorithm ◽  
Analysis Method ◽  
Fisher Linear Discriminant ◽  
Linear Discriminant ◽  
Discriminant Analysis Method ◽  
Pca Method ◽  
Better Than

This paper firstly analyzes the principle of face recognition algorithm, studies feature selection and distance criterion problem, puts forward the defects of PCA face recognition algorithm and LDA face recognition algorithm. According to the deficiencies and shortcomings of PCA face recognition algorithm and LDA face recognition algorithm, this paper proposes a solution -- PCA+LDA. The method uses the PCA method to reduce the dimensionality of feature space, it uses Fisher linear discriminant analysis method to classification, the realization of face recognition. Experiments show that, this method can not only improve the feature extraction speed, but also the recognition rate is better than single PCA method and LDA method.

scholarly journals A Semi-Supervised Transfer Learning with Grid Segmentation for Outdoor Localization over LoRaWans

Sensors ◽  
2021 ◽  
Vol 21 (8) ◽  
pp. 2640
Author(s):  
Yuh-Shyan Chen ◽  
Chih-Shun Hsu ◽  
Chan-Yin Huang
Keyword(s):  
Transfer Learning ◽  
Outdoor Environment ◽  
Localization Algorithm ◽  
Target Domain ◽  
Location Accuracy ◽  
Localization Scheme ◽  
Signal Features ◽  
Unlabelled Data ◽  
Outdoor Localization ◽  
Relationship Of

During the training phase of the supervised learning, it is not feasible to collect all the datasets of labelled data in an outdoor environment for the localization problem. The semi-supervised transfer learning is consequently used to pre-train a small number of labelled data from the source domain to generate a kernel knowledge for the target domain. The kernel knowledge is transferred to a target domain to transfer some unlabelled data into the virtual labelled data. In this paper, we have proposed a new outdoor localization scheme using a semi-supervised transfer learning for LoRaWANs. In the proposed localization algorithm, a grid segmentation concept is proposed so as to generate a number of virtual labelled data through learning by constructing the relationship of labelled and unlabelled data. The labelled-unlabelled data relationship is repeatedly fine-tuned by correctly adding some more virtual labelled data. Basically, the more the virtual labelled data are added, the higher the location accuracy will be obtained. In the real implementation, three types of signal features, RSSI, SNR, and timestamps, are used for training to improve the location accuracy. The experimental results illustrate that the proposed scheme can improve the location accuracy and reduce the localization error as opposed to the existing outdoor localization schemes.

scholarly journals Transfer Learning for LiDAR-Based Lane Marking Detection and Intensity Profile Generation

Geomatics ◽  
2021 ◽  
Vol 1 (2) ◽  
pp. 287-309
Author(s):  
Ankit Patel ◽  
Yi-Ting Cheng ◽  
Radhika Ravi ◽  
Yi-Chun Lin ◽  
Darcy Bullock ◽  
...  
Keyword(s):  
Transfer Learning ◽  
Fine Tuning ◽  
Target Domain ◽  
Independent Dataset ◽  
Lane Marking ◽  
Intensity Information ◽  
Training Examples ◽  
Learning Frameworks ◽  
Mobile Mapping Systems ◽  
Better Than

Recently, light detection and ranging (LiDAR)-based mobile mapping systems (MMS) have been utilized for extracting lane markings using deep learning frameworks. However, huge datasets are required for training neural networks. Furthermore, with accurate lane markings being detected utilizing LiDAR data, an algorithm for automatically reporting their intensity information is beneficial for identifying worn-out or missing lane markings. In this paper, a transfer learning approach based on fine-tuning of a pretrained U-net model for lane marking extraction and a strategy for generating intensity profiles using the extracted results are presented. Starting from a pretrained model, a new model can be trained better and faster to make predictions on a target domain dataset with only a few training examples. An original U-net model trained on two-lane highways (source domain dataset) was fine-tuned to make accurate predictions on datasets with one-lane highway patterns (target domain dataset). Specifically, encoder- and decoder-trained U-net models are presented wherein, during retraining of the former, only weights in the encoder path of U-net were allowed to change with decoder weights frozen and vice versa for the latter. On the test data (target domain), the encoder-trained model (F1-score: 86.9%) outperformed the decoder-trained (F1-score: 82.1%). Additionally, on an independent dataset, the encoder-trained one (F1-score: 90.1%) performed better than the decoder-trained one (F1-score: 83.2%). Lastly, on the basis of lane marking results obtained from the encoder-trained U-net, intensity profiles were generated. Such profiles can be used to identify lane marking gaps and investigate their cause through RGB imagery visualization.

scholarly journals A Two-Phase Fashion Apparel Detection Method Based on YOLOv4

2021 ◽  
Vol 11 (9) ◽  
pp. 3782
Author(s):  
Chu-Hui Lee ◽  
Chen-Wei Lin
Keyword(s):  
Object Detection ◽  
Transfer Learning ◽  
Detection Method ◽  
Phase Transfer ◽  
Recognition Task ◽  
Phase Detection ◽  
Target Domain ◽  
Two Phase ◽  
Detection Technology ◽  
Fashion Apparel

Object detection is one of the important technologies in the field of computer vision. In the area of fashion apparel, object detection technology has various applications, such as apparel recognition, apparel detection, fashion recommendation, and online search. The recognition task is difficult for a computer because fashion apparel images have different characteristics of clothing appearance and material. Currently, fast and accurate object detection is the most important goal in this field. In this study, we proposed a two-phase fashion apparel detection method named YOLOv4-TPD (YOLOv4 Two-Phase Detection), based on the YOLOv4 algorithm, to address this challenge. The target categories for model detection were divided into the jacket, top, pants, skirt, and bag. According to the definition of inductive transfer learning, the purpose was to transfer the knowledge from the source domain to the target domain that could improve the effect of tasks in the target domain. Therefore, we used the two-phase training method to implement the transfer learning. Finally, the experimental results showed that the mAP of our model was better than the original YOLOv4 model through the two-phase transfer learning. The proposed model has multiple potential applications, such as an automatic labeling system, style retrieval, and similarity detection.

scholarly journals Utilization of a Gas-Sensing System to Discriminate Smell and to Monitor Fermentation during the Manufacture of Oolong Tea Leaves

Micromachines ◽  
2021 ◽  
Vol 12 (1) ◽  
pp. 93
Author(s):  
Ting-Shiang Tseng ◽  
Mei-Hui Hsiao ◽  
Po-An Chen ◽  
Shu-Yen Lin ◽  
Shih-Wen Chiu ◽  
...  
Keyword(s):  
Gas Sensing ◽  
Critical Factor ◽  
Current Method ◽  
Sensory Function ◽  
Oxide Semiconductor ◽  
Tea Leaves ◽  
Linear Discriminant ◽  
Oolong Tea ◽  
Olfactory Sense ◽  
On Line

The operational duration of shaking tea leaves is a critical factor in the manufacture of oolong tea; this duration influences the formation of its flavor and fragrance. The current method to control the duration of fermentation relies on the olfactory sense of tea masters; they monitor the entire process through their olfactory sense, and their experience decides the duration of shaking and setting. Because of this human factor and olfactory fatigue, it is difficult to define an optimum duration of shaking and setting; an inappropriate duration of shaking and setting deteriorates the quality of the tea. In this study, we used metal-oxide-semiconductor gas sensors to establish an electronic nose (E-nose) system and tested its feasibility. This research was divided into two experiments: distinguishing samples at various stages and an on-line experiment. The samples of tea leaves at various stages exhibited large differences in the level of grassy smell. From the experience of practitioners and from previous research, the samples could be categorized into three groups: before the first shaking (BS1), before the shaking group, and after the shaking group. We input the experimental results into a linear discriminant analysis to decrease the dimensions and to classify the samples into various groups. The results show that the smell can also be categorized into three groups. After distinguishing the samples with large differences, we conducted an on-line experiment in a tea factory and tried to monitor the smell variation during the manufacturing process. The results from the E-nose were similar to those of the sense of practitioners, which means that an E-nose has the possibility to replace the sensory function of practitioners in the future.

scholarly journals Ripeness Prediction of Postharvest Kiwifruit Using a MOS E-Nose Combined with Chemometrics

Sensors ◽  
2019 ◽  
Vol 19 (2) ◽  
pp. 419 ◽  
Author(s):  
Dongdong Du ◽  
Jun Wang ◽  
Bo Wang ◽  
Luyi Zhu ◽  
Xuezhen Hong
Keyword(s):  
Prediction Models ◽  
Extraction Methods ◽  
Oxide Semiconductor ◽  
Soluble Solids ◽  
Support Vector ◽  
Least Squares Regression ◽  
Accuracy Rate ◽  
Linear Discriminant ◽  
The Difference ◽  
Ripe Stage

Postharvest kiwifruit continues to ripen for a period until it reaches the optimal “eating ripe” stage. Without damaging the fruit, it is very difficult to identify the ripeness of postharvest kiwifruit by conventional means. In this study, an electronic nose (E-nose) with 10 metal oxide semiconductor (MOS) gas sensors was used to predict the ripeness of postharvest kiwifruit. Three different feature extraction methods (the max/min values, the difference values and the 70th s values) were employed to discriminate kiwifruit at different ripening times by linear discriminant analysis (LDA), and results showed that the 70th s values method had the best performance in discriminating kiwifruit at different ripening stages, obtaining a 100% original accuracy rate and a 99.4% cross-validation accuracy rate. Partial least squares regression (PLSR), support vector machine (SVM) and random forest (RF) were employed to build prediction models for overall ripeness, soluble solids content (SSC) and firmness. The regression results showed that the RF algorithm had the best performance in predicting the ripeness indexes of postharvest kiwifruit compared with PLSR and SVM, which illustrated that the E-nose data had high correlations with overall ripeness (training: R2 = 0.9928; testing: R2 = 0.9928), SSC (training: R2 = 0.9749; testing: R2 = 0.9143) and firmness (training: R2 = 0.9814; testing: R2 = 0.9290). This study demonstrated that E-nose could be a comprehensive approach to predict the ripeness of postharvest kiwifruit through aroma volatiles.

P-121: Objective assessment of mild cognitive impairment by Fisher linear discriminant

2007 ◽  
Vol 3 (3S_Part_2) ◽  
pp. S137-S137
Author(s):  
Zhilin Wu ◽  
Piero Antuono ◽  
Guofan Xu ◽  
Jennifer Jones ◽  
Shi-jiang Li
Keyword(s):  
Cognitive Impairment ◽  
Mild Cognitive Impairment ◽  
Objective Assessment ◽  
Fisher Linear Discriminant ◽  
Linear Discriminant
Sign in / Sign up

Export Citation Format

Share Document