SMART ANALYSIS OF AUTOMATED AND SEMI-AUTOMATED APPROACHES TO DATA ANNOTATION FOR MACHINE LEARNING
Abstract
Data annotation for machine learning is the process of labeling data so that machines can properly identify patterns and other related information. It is a critical task within many artificial intelligence (AI) and machine learning (ML) projects. The traditional approach to data annotation involves manual input from a knowledgeable human expert. This, however, can be extremely costly, both in terms of time and money. To help reduce these costs, automated and semi-automated approaches to data annotation have been explored. Automated approaches are computer programs that label data automatically without any human input. However, there are issues with automated techniques such as potential errors, bias, and uncertainty. Semi-automated approaches are gaining popularity because they involve less manual labor while still allowing a human expert to verify the output of the program. Some of the more popular semi-automated approaches include machine teaching, rule-based systems, and active learning. Machine teaching is an approach to data annotation that is based on reinforcement learning. Through the use of reinforcement learning, a human user provides feedback to an annotating system, and the system uses this feedback to learn to improve.

Authors
M. Sutharsan
Selvam College of Technology, India

Keywords
Data, Annotation, Machine Learning, Artificial Intelligence, Teaching, Rules-Based Systems
Yearly Full Views
JanuaryFebruaryMarchAprilMayJuneJulyAugustSeptemberOctoberNovemberDecember
000010000000
Published By :
ICTACT
Published In :
ICTACT Journal on Data Science and Machine Learning
( Volume: 4 , Issue: 3 , Pages: 467 - 470 )
Date of Publication :
June 2023
Page Views :
145
Full Text Views :
8

Creative Commons License
This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.