Pushpendra Kumar

Dr. Pushpendra Kumar

orcid

Research Focus & Metrics

Representative Publications

View All Publications
7 April 2018 in International Journal of Information Technology
Verified
Journal Article
Closed access
PDF file
PKPushpendra Kumar
Pushpendra Kumar
RTRamjeevan Singh Thakur
Ramjeevan Singh Thakur

22 January 2020 in Multimedia Tools and Applications
Verified
Journal Article
Closed access
PDF file
PKPushpendra Kumar
Pushpendra Kumar
RTRamjeevan Singh Thakur
Ramjeevan Singh Thakur

15 July 2024 in IEEE Access
Verified
Journal Article
Open access
MPMuskan Priyadarshani
Muskan Priyadarshani
PKPushpendra Kumar
Pushpendra Kumar
KBKanojia Sindhuben Babulal
Kanojia Sindhuben Babulal
DRDharmendra Singh Rajput
Dharmendra Singh Rajput
HPHarshita Patel
Harshita Patel
Emotion Recognition is a critical area of research including healthcare, human-computer interaction, and psychology. While traditional methods mainly rely on facial expressions and textual analysis, they also have inherent flaws and cannot be reliable. Facial expression-based emotion recognition assumes that it represents genuine internal emotions that may be inaccurate. Similarly, textual analysis depends on the available data and needs help accurately capturing subtle emotions in text. However, electroencephalography (EEG) has emerged as a rising alternative for objective and real-time emotion recognition. Unlike facial and textual methods, EEG directly measures brain activity and provides a reliable result. To address this researchers have used basic machine learning methods that need manual feature extraction, which might miss essential data and make the process slow and less accurate. In this study, we propose a comprehensive methodology for EEG-based emotion recognition that addresses the limitations of traditional methods and basic machine learning techniques. Our approach involves preprocessing EEG signals using a butter-worth bandpass filter to eliminate noise, followed by feature extraction techniques. We then employ Principal Component Analysis (PCA) for dimensionality reduction, ensuring efficient data representation. To further enhance the model performance we explore machine learning classifiers(GaussianNB, SVM, Random Forest) and proposed an EEG-LSTM and GRU model with an accuracy of 97% and 96% respectively, that gives better results than the basic machine learning models.

1 January 2021 in Computers, Materials & Continua
Verified
Journal Article
Open access
PKPushpendra Kumar
Pushpendra Kumar
RTRamjeevan Singh Thakur
Ramjeevan Singh Thakur
The aim of this research is to develop a mechanism to help medical practitioners predict and diagnose liver disease. Several systems have been proposed to help medical experts by diminishing error and increasing accuracy in diagnosing and predicting diseases. Among many existing methods, a few have considered the class imbalance issues of liver disorder datasets. As all the samples of liver disorder datasets are not useful, they do not contribute to learning about classifiers. A few samples might be redundant, which can increase the computational cost and affect the performance of the classifier. In this paper, a model has been proposed that combines noise filter, fuzzy sets, and boosting techniques (NFFBTs) for liver disease prediction. Firstly, the noise filter (NF) eliminates the outliers from the minority class and removes the outlier and redundant pair from the majority class. Secondly, the fuzzy set concept is applied to handle uncertainty in datasets. Thirdly, the AdaBoost boosting algorithm is trained with several learners viz, random forest (RF), support vector machine (SVM), logistic regression (LR), and naive Bayes (NB). The proposed NFFBT prediction system was applied to two datasets (i.e., ILPD and MPRLPD) and found that AdaBoost with RF yielded 90.65% and 98.95% accuracy and F1 scores of 92.09% and 99.24% over ILPD and MPRLPD datasets, respectively.

8 December 2018 in Iran Journal of Computer Science
Verified
Journal Article
Open access
PDF file
PKPushpendra Kumar
Pushpendra Kumar
VKVinod Kumar
Vinod Kumar
RTRamjeevan Singh Thakur
Ramjeevan Singh Thakur
Recommendation systems are most commonly used to recommend items for web users. It assists users in the selection of product from millions of product. E-Commerce websites such as AMAZON recommend items to its customers. The recommendation system mainly depends upon the previous history of its users. In this paper, a new User Rating Prediction (URP) algorithm is proposed to predict ratings for items. The proposed URP algorithm mainly depends upon similarity of users and assumes that users with similar taste may be interested in similar items. The proposed system first makes a list of related users for every user and then uses this information to predict ratings for different items. The result of the proposed algorithm was compared with the previous existing methods. The proposed algorithm gives small value of Mean Absolute Error (MAE) and root-mean-square error (RMSE) as compared to other methods.

1 March 2019
Verified
Conference Paper
Closed access
PKPushpendra Kumar
Pushpendra Kumar
RTRamjeevan Singh Thakur
Ramjeevan Singh Thakur
In today's world, the liver disorder is a serious disease and the number of people with liver disease continues to increase. The side effect of the disease is hard to identify by the medical expert because people experience sound even they are suffering from the disease. Classification techniques are broadly applied in the prediction of liver disease from a liver function test (LFT) datasets. Issues with many real-world datasets including liver disease data are a class imbalance. The traditional classification techniques produce biased results towards the majority classes. In this article, a novel Fuzzy-ANWKNN approach is proposed by hybridization of Fuzzy based Adaptive K-NN and Neighbor Weighted K-NN to efficiently predict the liver disease. The performance of the Fuzzy-ANWKNN is found encouraging as compared with Fuzzy-ADPTKNN method using ILPD and MPRLPD imbalance dataset. The experimental evaluation of the proposed Fuzzy-ANWKNN method confirms that the proposed model is better than the existing Fuzzy-ADPTKNN method.