• D P Gaikwad

Sentiment analysis is a classification problem which is used to predict the polarity of words, paragraph or documents and then classify them into positive or negative sentiment. Sentiment analysis offers people a fast and effective way to measure the feelings towards product, their party and films etc. The primary issue in sentiment analysis techniques is the determination of the most appropriate classifier for a given classification problem. Several machine-learning techniques such as logistic regression, support vector machine and Naive Bayes have used to classify feelings of people. Now a day, deep learning is mostly used for sentiment analysis. In this paper, we propose the N-gram evaluation based Multilayer Perceptron as a sentiment analyser. The movie review dataset have been used for training and testing Multilayer Perceptron with different parameter. For a given dataset, the goal is to find the optimal parameters of Multilayer Perceptron to achieve maximum accuracy while minimizing computation time required for training and testing. In this experimental work, Multilayer Perceptron have trained and tested using different hyper parameters and different optimization functions. The experimental results show that the proposed N-gram evaluation based Multilayer Perceptron exhibits best training accuracy with minimum training time using Adam optimization function. It is also found that both RMSprop and Adam exhibits approximately same accuracy on test dataset.

N-gram, Machine Learning, Adam, Multilayer Perceptron, Deep learning, Dropout rate.

"PERFORMANCE OF MULTILAYER PERCEPTRON BASED SENTIMENT ANALYSIS ON HYPER PARAMETERS AND OPTIMIZERS", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.6, Issue 5, page no.408-416, May-2019, Available :http://www.jetir.org/papers/JETIR1905Q58.pdf

"PERFORMANCE OF MULTILAYER PERCEPTRON BASED SENTIMENT ANALYSIS ON HYPER PARAMETERS AND OPTIMIZERS", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.6, Issue 5, page no. pp408-416, May-2019, Available at : http://www.jetir.org/papers/JETIR1905Q58.pdf