Robust Classification with Noisy Labels for Manufacturing Applications: A Hybrid Approach Based on Active Learning and Data Cleaning

2021 ◽  
Author(s):  
Shuo Zhao ◽  
Xin Li ◽  
Ying-Chi Chen
Keyword(s):  
Active Learning ◽  
Data Cleaning ◽  
Hybrid Approach ◽  
Robust Classification ◽  
Manufacturing Applications ◽  
Noisy Labels

scholarly journals Drift Compensation on Massive Online Electronic-Nose Responses

Chemosensors ◽  
2021 ◽  
Vol 9 (4) ◽  
pp. 78
Author(s):  
Jianhua Cao ◽  
Tao Liu ◽  
Jianjun Chen ◽  
Tao Yang ◽  
Xiuxiu Zhu ◽  
...  
Keyword(s):  
Active Learning ◽  
Gas Sensor ◽  
Electronic Nose ◽  
Effective Means ◽  
Learning Paradigm ◽  
Class Label ◽  
Learning Framework ◽  
Sensor Drift ◽  
Drift Compensation ◽  
Noisy Labels

Gas sensor drift is an important issue of electronic nose (E-nose) systems. This study follows this concern under the condition that requires an instant drift compensation with massive online E-nose responses. Recently, an active learning paradigm has been introduced to such condition. However, it does not consider the “noisy label” problem caused by the unreliability of its labeling process in real applications. Thus, we have proposed a class-label appraisal methodology and associated active learning framework to assess and correct the noisy labels. To evaluate the performance of the proposed methodologies, we used the datasets from two E-nose systems. The experimental results show that the proposed methodology helps the E-noses achieve higher accuracy with lower computation than the reference methods do. Finally, we can conclude that the proposed class-label appraisal mechanism is an effective means of enhancing the robustness of active learning-based E-nose drift compensation.

scholarly journals Agreeing to disagree: active learning with noisy labels without crowdsourcing

2017 ◽  
Vol 9 (8) ◽  
pp. 1307-1319 ◽  
Author(s):  
Mohamed-Rafik Bouguelia ◽  
Slawomir Nowaczyk ◽  
K. C. Santosh ◽  
Antanas Verikas
Keyword(s):  
Active Learning ◽  
Agreeing To Disagree ◽  
Noisy Labels

A Hybrid Approach to Feature Recognition Using Approximate and Partial Exact CAD Models

2002 ◽  
Author(s):  
Nagesh Belludi ◽  
Derek Yip-Hoi
Keyword(s):  
Feature Recognition ◽  
Hybrid Approach ◽  
Parameter Extraction ◽  
Feature Identification ◽  
Cad System ◽  
Processing Times ◽  
File Formats ◽  
Cad Models ◽  
Manufacturing Applications ◽  
Simple Surface

Several CAD system independent feature recognition techniques have been developed to drive manufacturing applications. Commercial implementations of these techniques require translating CAD models using STEP or other neutral file formats. With large CAD models found in some application domains; e.g., powertrain machining, corresponding STEP files are also large. This leads to large processing times. Another approach is to use lightweight formats such as STL or VRML. Here, complete & accurate parameter extraction is difficult because these formats approximate surfaces as tessellations. This paper discusses a new methodology for feature recognition, in which a VRML file is used for feature identification. To some extent, parameters of faces with simple surface-types are recovered from the tessellated model. If identified features consist of faces whose parameters are not recovered from the tessellated model, a partial STEP file translation is used for extracting exact parameters. This CAD system independent algorithmic development and implementation reduces the amount of data exported to neutral files, thus leading to more efficient feature recognition.

scholarly journals Robust Classification from Noisy Labels: Integrating Additional Knowledge for Chest Radiography Abnormality Assessment

2021 ◽  
pp. 102087
Author(s):  
Sebastian Gündel ◽  
Arnaud A.A. Setio ◽  
Florin C. Ghesu ◽  
Sasa Grbic ◽  
Bogdan Georgescu ◽  
...  
Keyword(s):  
Chest Radiography ◽  
Robust Classification ◽  
Noisy Labels

A hybrid approach to active learning

2001 ◽  
Vol 36 (2) ◽  
pp. 124-128 ◽  
Author(s):  
R D Ramsier
Keyword(s):  
Active Learning ◽  
Hybrid Approach
Sign in / Sign up

Export Citation Format

Share Document