Dr. Amit Vajpayee
@cuchd.in
Professor in Computer Science and Engineering
Chandigarh University
RESEARCH, TEACHING, or OTHER INTERESTS
Computer Science, Artificial Intelligence, Computer Vision and Pattern Recognition, Neuroscience
Scopus Publications
Scopus Publications
Amit Vajpayee, Palak Preet Kaur, Neeraj Bali, Ankit Sharma, Alankrita Aggarwal, and Abhineet Anand
AIP Publishing
Amit Vajpayee, Sulekha Singh, Neeraj Bali, Ankit Sharma, Alankrita Aggarwal, and Abhineet Anand
AIP Publishing
Amit Vajpayee, Palak Preet Kaur, Neeraj Bali, Ankit Sharma, Alankrita Aggarwal, and Abhineet Anand
AIP Publishing
Yash Suthar, Chintan Thacker, and Amit Vajpayee
Springer Nature Singapore
Vijay Bharatbhai Vasaiya, Amit Vajpayee, and Ankita Gandhi
IEEE
Wheat is a vital crop for global food security, yet its production faces significant threats from various diseases, leading to substantial yield losses. Traditional detection methods, such as manual scouting and laboratory analysis, are often labour-intensive and inefficient. The advent of deep learning techniques has revolutionized agricultural disease detection, offering more efficient and accurate solutions. We provide a comprehensive overview of common wheat diseases, methodologies, and datasets utilized in recent studies. A critical analysis of the literature reveals performance metrics of various deep learning models compared to traditional methods, Furthermore, the review suggests future research directions, emphasizing the integration of IoT technologies for real-time monitoring and the use of advanced sensors for improved detection accuracy. This synthesis aims to enhance strategies for effective disease management in wheat cultivation.
Santosh Varshney, Amit Vajpayee, Bhupinder Kaur, and Poonam Kukana
IEEE
Santosh Varshney, Amit Vajpayee, Poonam Kukana, and Bhupinder Kaur
IEEE
Praveen Kumar Patidar, Navdeep Kaur, Amit Vajpayee, Pradeepta Kumar Sarangi, and Alok Kumar Agrawal
IEEE
Recent advancements in machine learning (ML) and artificial intelligence (AI) have greatly enhanced the detection and diagnosis of healthcare issues, including chronic kidney disease (CKD). CKD, also known as chronic kidney failure, involves the gradual deterioration of kidney function, which is essential for filtering waste and excess fluids from the blood. As the disease progresses, harmful fluids, electrolytes, and toxins can accumulate, posing serious health risks. The term "chronic" indicates the slow progression of this condition, which is seeing rising incidence globally. Patients with severely impaired kidney function typically survive without intervention for only about 18 days, resulting in urgent demand for kidney transplants and dialysis treatments. In addressing CKD, this study employs seven different ML algorithms to analyze kidney disease data and identify the most effective model. Many popular models such as Random Forest and Decision Tree models have been proved as useful tools. This study evaluates six distinct machine learning algorithms to analyze kidney disease data and identify the most effective model. The findings indicate that the LG Boost model achieved an impressive accuracy rate of 98.3%. Additionally, Stochastic Gradient Boosting and XGBoost models also have performed very well with an impressive accuracy of 97.5%.
Alok Kumar Agrawal, Amit Vajpayee, Merry Saxena, Pradeepta Kumar Sarangi, Karan Bajaj, and Ashok Kumar Sahoo
Springer Nature Switzerland
Amit Vajpayee, Abhineet Anand, Ankit Sharma, Palakpreet Kaur, Jaspreet Singh, and Amit Verma
CRC Press
Shweta Agarwal, Neetu Rani, and Amit Vajpayee
Wiley
Amit Vajpayee, Palak Preet Kaur, Ankit Sharma, and Abhishek Tiwari
IEEE
Pattern recognition and classification standards are important for data interpretation and decision-making processes in many fields. This article provides a comprehensive review of these two intersections, exploring their evolution, approaches, recent developments, and practical applications. We take a closer look at various applications in information modelling and classification, highlighting their advantages and limitations. We discuss simple processes such as supervised and unsupervised learning, emphasizing the role of features in effective representation. Classification algorithms their advantages, and limitations are described. The review continues to discuss the intersection of these areas and shows how they complement each other in solving problems. We also explore their diverse uses, from medical to autonomous driving, to demonstrate their transformative potential. This review aims to provide insight into recognition and classification standards, encouraging further research and innovation in these important areas.
Amit Vajpayee, Palak Preet Kaur, Ankit Sharma, and Santosh Varshney
IEEE
Neuromorphic computing is a new paradigm that emerges from the structure and function of the human brain and aims to revolutionize computing. The technology is designed to simulate the high speed, low power consumption and large contraction of the brain. Neuromorphic systems provide great potential for intelligence and big data by using special algorithms and special designs to simulate neural networks. Despite encouraging progress from technology companies such as Intel and IBM, the development of neural computing chips still faces major challenges. This approach brings significant benefits to artificial intelligence, machine learning and big data. This technology is still in its infancy but has great potential to revolutionize intelligence. By mimicking the brain's performance and learning potential, neuromorphic computing could pave the way for smarter, faster, and more flexible systems that work well on water, performing tasks such as complex data analysis and on-the-fly decision making.
Amit Vajpayee, Palak Preet Kaur, Ankit Sharma, and Sakshi
IEEE
Finding comparable photos using a query is known as image retrieval, and it has proven essential to many applications. This review explores the development of computer vision methods that allow for the effective retrieval of visual content. We examine the groundwork established by feature-based methods such as SIFT, which completely changed the field of image retrieval. We explore methods such as hierarchical deep metric learning and contextual pooling, demonstrating how they might take use of spatial linkages to increase retrieval accuracy. The research also investigates the potential of Generative Adversarial Networks (GANs) for image retrieval tasks by allowing picture alteration for more sophisticated searches. We may learn a great deal about the state-of-the-art and set the path for future directions in computer vision and image retrieval research by looking at these developments.
Apoorva Jindal, Riya Kamboj, Sakshi Pathak, Kunal Dubey, and Amit Vajpayee
IEEE
Within the sphere of healthcare, the widespread employment of large datasets has infiltrated each element of the field, starting from groundbreaking scientific inquiry right through to optimizing patient interactions and therapeutic results. With a myriad of diverse ailments challenging healthcare systems worldwide, the integration of Machine Learning and Big Data technologies has emerged as a novel approach to disease prediction and diagnosis. This research embarks on a transformative journey, investigating the application of machine learning algorithms to forecast diseases based on presenting symptoms and provide drug recommendation. Through the implementation of Naive Bayes, Decision Trees, Random Forests, and Logistic Regression, this study explores the path to more accurate and data-driven healthcare solutions.
Farida A. Ali, Baibaswata Mohapatra, Amit Vajpayee, Pradeepta Kumar Sarangi, Srikanta Kumar Mohapatra, and Alok Kumar Agrawal
IEEE
Now adays, heart disease is one of the major problems in human life. This disease alone is responsible for dying ofmore people per year in comparison to all the other diseases integrated. It includes various problems related to the heart like Blood vessel disease (coronary artery disease), Born heart problems (congenital heart defects), Irregular heartbeats(arrhythmias) and many more. Also, this disease is very commonly seen in this new generation especially the youths. Study says youths are suffering with heart diseases due to chronic stress, smoking, poor lifestyle, etc. So, the growing cases of heart disease makes us feel that we should have some advanced algorithms and works regarding heart disease prediction. This work intends to find whether the patient or the person is suffering or likely to suffer from heart disease or not. This research paper aims to implement supervised machine learning like Decision trees (DT), Random Forest (RF), K-nearest neighbors, Gaussian Naï ve Bayes algorithms, Logistic Regression and SVC. The dataset used in this work has been collected from Kaggle and the accuracy by Decision Tree and Random Forest are 98.33%.
Ramamani Tripathy, Dibyahash Bordoloi, Amit Vajpayee, Merry Saxena, Srikanta Kumar Mohapatra, and Pradeepta Kumar Sarangi
IEEE
Diabetes is a constant sickness that creates when the pancreas neglects to produce sufficient chemical to manage blood glucose levels, or when the body can't utilize insulin as expected. One of the main characteristics of diabetes, a class of metabolic illnesses, is hyperglycemia. Type 1 diabetes is a hereditary disorder that often appears early in life and has a lengthy incubation period. People with type 2 diabetes have improper insulin sensitivity in their cells. It evolves with time and primarily depends on individual lifestyle choices. Diabetes is the most frequent reason for damage and dysfunction of organs and tissue, which includes heart failure, stroke, kidney failure, and blindness. Early diabetes detection is so essential. This work implements and compares the effectiveness of four machine learning models. The first model implemented is XGBoost Classifier with accuracy obtained as 91%, the second model implemented is Random Forest (RF) with reported accuracy as 90% and the third model is Decision Tree with accuracy 88% and the fourth one is Support Vector Classification (SVC) with reported accuracy as 88%.
Pradhyuman Singh Shaktawat, Ritika Thakur, Nishit Hirani, Prabhat Panda, Surjeet Surjeet, and Amit Vajpayee
IEEE
This study examines the concerning rise of alcohol consumption among high school students and its negative consequences on academic achievement. We explore the influence of socio-educational factors, including socioeconomic background, peer relationships, and school environment, on alcohol use patterns. Additionally, we investigate the moderating role of parental engagement and family dynamics in the relationship between alcohol consumption and academic performance. Utilizing the Student Alcoholism and Academic Performance Dataset, we employ data preprocessing techniques, we address missing data, convert categorical variables into a suitable format for modelling, and normalize numerical features to prevent bias. Through exploratory data analysis, we gain insights into descriptive statistics, data distributions, and correlations, revealing the extent of alcohol's impact on academic outcomes. Dimensionality reduction via Principal Component Analysis (PCA) streamlines the dataset for further analysis. We implement the Random Forest Regression algorithm to develop predictive models for academic performance based on the identified socio-educational factors and alcohol consumption habits. Evaluating model performance using mean squared error calculations, we aim to provide actionable insights for designing targeted interventions. These interventions will focus on mitigating the detrimental effects of adolescent alcohol use on student success and promoting a healthier learning environment.
Pradeepta Kumar Sarangi, Shreya Kumari, Mani Sawhney, Amit Vajpayee, Mukesh Rohra, and Srikanta Mallik
IGI Global
The digital world is replete with data like cyber security data, internet of things (IoT) data, enterprise data, mobile data, health data, and more. To analyse this data brilliantly and develop intelligent and automated applications, everyone has to know artificial intelligence (AI) algorithms, deep learning (DL) and machine learning (ML). Therefore, in today's technology-driven or digital world, no company can afford to ignore artificial intelligence or machine learning. Machine learning is a subfield of artificial intelligence, which is the scientific study of algorithms and statistical models that a computer system utilises to effectively carry out a given task without the need for any explicit instructions. This chapter begins with the basics of machine learning and its diverse range of techniques. This chapter also discusses various classification and clustering methods along with their applications and concludes with some real-world applications and examples and research development using machine learning and quantum computing in healthcare.
Vinay Gautam, Amit Vajpee, and Abhishek
AIP Publishing
Sunil Gupta, Rakesh Saxena, Ankit Bansal, Kamal Saluja, Amit Vajpayee, and Shikha
AIP Publishing
Gurpreet Singh Panesar, Amit Vajpayee, and Nikhil Agarwal
IEEE
The image's characteristics and density chart serve as the foundation for the novel object detection described in this study. The two key stages of the suggested technique are object localization and bounding box estimation. To precisely locate items in an image, object localization makes use of the spatial distribution of objects learned from the density map. Bounding box estimation calculates the edges of the observed items using clustering and common edge methods. The scale variation brought on by unclear perspective is one of the key difficulties in object density map estimation. The suggested solution incorporates a novel technique for calculating the previous focus map for each image in order to overcome this problem. The focus power of this method is based on sparse defocus dictionary learning on a newly created dataset, and it takes into account the quantity of non-zero dictionary atom coefficients. Unlike existing edge density estimation approaches, the proposed framework captures spatial features and allows for threshold type selection in various ways. The proposed methodology offers promising results for object detection tasks and has the potential to improve accuracy in various image analysis applications.
Rahul Bhandari, Amit Vajpayee, Ravi Kumar, and Daulat Sihag
IEEE
Motion estimation algorithm is most important and curial component in compressing a video. Block matching search pattern is one of the popular algorithm is used in motion estimation. A proposed searching pattern was developed in this research work to employ in collaboration with a fast block-based motion estimation technique. The proposed searching pattern, as compared to the preceding patterns employed in the block motion estimation approach, may produce motion vectors with fewer search points and hence requires less search points. The suggested searching pattern outperforms the existing technique, especially when it comes to finding huge motion vectors. However, when compared to several previously connected approaches, this study reveals that the suggested pattern decreases search spots while boosting the accuracy of motion vectors.
Sarita Simaiya, Umesh Kumar Lilhore, Ranjan Walia, Shweta Chauhan, and Amit Vajpayee
IEEE
Brain disease is the most severe, pervasive, and life-threatening illness globally. Globally, brain tumours destroy the lifestyles of numerous individuals annually due to the fast expansion of tumour tissue. To prevent the deaths of individuals worldwide, prompt evaluation and classification of brain diagnosis are therefore necessary. In past years, the three primary forms of brain tumours, gliomas, malignant tumours, and pituitary, have been detected and classified most frequently using deep learning techniques. However, the volume of the sample with annotations significantly impacts how well deep learning algorithms work. It becomes complicated to classify a massive volume of medical data. In this work, we offer a stable hybrid approach for brain tumour identification predicated using CNN (VGG-16) with transfer learning, which incorporates the classic confusion and uncertainty polling technique, a best-fit methodology pooled by best-fit methodology using VGG-16, as well as the transfer learning approach. This approach decreases tagging expenditures while preserving the resilience and consistency of the systems. This research utilises the online brain tumour Kaggle dataset. To compare the proposed dynamic transfer learning strategy with the current CNN technique, evaluation measures such as precision, f-measure, and accuracy are used. The proposed technique enhances the CNN strategy by 5%, achieving 96.77% accuracy, 98.7% recall, 96.35 precision, and 96.78% F-measure.
Anuj Kumar Jain, Vikrant Sharma, Sandeep Goel, Raj Gaurang Tiwari, Amit Vajpayee, and Rahul Bhandari
IEEE
The face, a crucial bodily feature, communicates a lot of information. Facial expressions, such as blinking and yawning more often than usual, differ from those in normal conditions when a driver is fatigued. A major contributing element in a significant proportion of auto accidents is driver inattentiveness. In recent years, this driver sleepiness detection method has received great praise and has been applied in several scenarios, such as visual attention monitoring and driver activity tracking. Studies show that inattentive and fatigued drivers are to blame for around 25% of severe highway fatal crashes, which is a lot more serious than drunk driving. In this paper, we suggest an algorithm that will see the face of the person who is driving and the eyes to be able to assess the degree of tiredness brought on by sluggish eye closure. Rapid prototyping of this participatory research and Innovation would be fruitful for sustainable communities.