International Journal of Innovative Technology and Exploring Engineering (IJITEE) ISSN: 2278-3075, Volume-8, Issue-9S3, July 2019 1517 Retrieval Number: I33160789S319/2019©BEIESP DOI: 10.35940/ijitee.I3316.0789S319 Published By: Blue Eyes Intelligence Engineering & Sciences Publication Handwriting Recognition by Machine Learning Anandika Sharma, Anupam Sharma Abstract— Handwriting is one of the most natural ways of communication among people. The handwriting recognition task is the main concern of scientific community because handwriting can be varies with the same person or from one person to another hence the prediction of human behavior through handwriting is a complex task. Earlier the handwriting analysis has been done by graphologists but due to the modernization and the arrival of digital world the handwriting analysis can be done with the help of computer aided machines. Different software and algorithms has been defined to do the analysis. In the new world of machine learning handwriting recognition and the prediction of human behavior can be done by using different techniques of machine learning which increase the speed of analysis This paper studies the recent advances and the trends in the field of handwriting recognition by machine learning. Keywords— handwriting, graphologist, machine learning. I. INTRODUCTION Handwriting is one of the most natural ways of communication among people. Every individual has different style of writing that is why people can judge the behavior of different people through handwriting. The handwriting recognition task is the main concern of scientific community because handwriting can be varies with the same person or from one person to another hence the prediction of human behavior through handwriting is a complex task. Manuscript Received on July 22, 2019 Anandika Sharma, School of humanities and social sciences Thapar institute of engineering and technology Patiala, Punjab Dr. Anupam Sharma, School of humanities and social sciences Thapar institute of engineering and technology Patiala, Punjab Earlier the handwriting analysis has been done by graphologists but due to the modernization and the arrival of digital world the handwriting analysis can be done with the help of computer aided machines. Different software and algorithms has been defined to do the analysis. In the new world of machine learning handwriting recognition and the prediction of human behavior can be done by using different techniques of machine learning which increase the speed of analysis. This paper studies the recent advances and the trends in the field of handwriting recognition by machine learning methods. Earlier the graphology method is used to evaluate and understand the handwriting pattern to predict behavior of human. But this method is time consuming and costly method. So later on the computer aided methods has been used to recognize the handwriting pattern and behavior of human. Handwriting analysis is one if the technique to predict the human personality including behavior, anger, self-esteem, honesty, fears and so on [1]. Different parameters of personality trait has been taken such as baseline, slant, margin, width and so on as the input to the machine learning methods and gives output in the form of personality trait. Detailed review of different authors has explained in this paper. II. OBJECTIVES 1. To predict the human behaviour through handwriting. 2. To study the recent advances and the trends in the field of handwriting recognition by machine learning techniques. III. SUMMARY OF RECENT ADVANCES AND TRENDS Prachi desai, aayush dhavale et al (2015) [2], depicts the graph-logical technique in which the human personality can be determine by evaluating the various feature from handwriting like page margin, slant of alphabet, the baseline etc. Here the computer aided methods can be used over graphological method to extract the feature of handwriting which helps to increase the speed and efficiency of analysis. Machine learning tool is used for the analysis and the implementation of machine learning can be based on the neural network in which baseline,