Ajay Kumar has a total of 29 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • HMG-Former: a hierarchical meta-graph transformer framework for robust multi-class SAR target recognition

    by Bibek Kumar, Ajay Kumar, Shashi Kant Gupta, Diaa Salama AbdElminaam, Deema Mohammed Alsekait, Mohammed Yahya Albeshri

  • Open Access

    SWARM INTELLIGENCE–OPTIMIZED DEEP LEARNING MODELS FOR DROUGHT PREDICTION AND SMART IRRIGATION SCHEDULING

    by Ajay Kumar, Tanu Gupta, Pragati Deogharkar, Rohaila Naaz, Chhavi Sharma

    Drought forecasting serves as a vital component for managing agricultural water resources in semi-arid areas, which experience climate-related changes. The research presents a deep learning system that uses swarm intelligence to forecast drought conditions and create irrigation plans for the Marathwada region of Maharashtra, India (18.0°N–20.5°N, 74.5°E–77.5°E). The research team constructed a multivariate time-series dataset by combining monthly hydro-meteorological data from the NASA POWER database with soil moisture and vegetation index data. The research team developed recurrent architectures through LSTM, GRU, and hybrid CNN-LSTM models, which they optimised through Particle Swarm Optimisation (PSO) and Grey Wolf Optimiser (GWO) methods. The optimised GWO-CNN-LSTM model demonstrated superior predictive capabilities with an R2 value of 0.837 and RMSE value of 0.422, which showed statistically significant enhancements compared to baseline models at p < 0.05. The five-fold time-series cross-validation process showed strong temporal reliability with an R2 value of 0.838 and a standard deviation of 0.007. The predictive outputs were integrated into an adaptive irrigation module, resulting in a 15% reduction in irrigation water usage and a 20.28% decrease in drought variance, indicating improved stability under dynamic climatic conditions. The predictive-prescriptive framework provides a data-driven, scalable solution for implementing climate-resilient precision agriculture and sustainable water resource management.

  • Open Access

    THUNDERSTORM CLASSIFICATION OVER KARNATAKA REGION USING CNN-LSTM HYBRID APPROACH

    by Hs Madhusudhan, Ajay Kumar, Punit Gupta

    Climate modeling is confronting a transformative shift due to incorporation of deep learning (DL) techniques. These methods, which are based on data, provide novel means to enhance the precision, efficiency, and comprehensibility of climate simulations and forecasts. Thunderstorms are one of the climate change issues which impact the economy and society. In this work we introduce a hybrid approach using CNN and LSTM to classify thunderstorm and non-thunderstorms using IMD dataset. Dataset consisting of observations from 1990 to 2021 is used for the classification. Combination of CNN-LSTM is compared with conventional machine learning techniques and result shows that CNN-LSTM outperforms ML techniques. CNN-LSTM method gives accuracy of 95.82% which has the highest performance compared to other techniques.

  • Open Access

    Early detection of tuberculosis using hybrid feature descriptors and deep learning network

    by Garima Verma, Ajay Kumar, Sushil Dixit

    Purpose: To detect tuberculosis (TB) at an early stage by analyzing chest X-ray images using a deep neural network, and to evaluate the efficacy of proposed model by comparing it with existing studies.Material and methods: For the study, an open-source X-ray images were used. Dataset consisted of two types of images, i.e., standard and tuberculosis. Total number of images in the dataset was 4,200, among which, 3,500 were normal chest X-rays, and the remaining 700 X-ray images were of tuberculosis patients. The study proposed and simulated a deep learning prediction model for early TB diagnosis by combining deep features with hand-engineered features. Gabor filter and Canny edge detection method were applied to enhance the performance and reduce computation cost.Results: The proposed model simulated two scenarios: without filter and edge detection techniques and only a pre-trained model with automatic feature extraction, and filter and edge detection techniques. The results achieved from both the models were 95.7% and 97.9%, respectively.Conclusions: The proposed study can assist in the detection if a radiologist is not available. Also, the model was tested with real-time images to examine the efficacy, and was better than other available models.


  • Retracted

    Retracted: An Integration Of Wireless Communications And Artificial Intelligence For Autonomous Vehicles

    by Kadambari Raghuram, Ibrahim Altarawni, Vijayalakshmi K, Ajay Kumar, Abhijeetsinh Jadeja, Pankaj Chandra

    The pinnacle of transportation is the development of autonomous driving, which, with the help of CAVs and related traffic management systems, can eventually lead to congestion- and accident-free driving. This vision has as of late prodded extraordinary examination interest in fields including IoV, LTE-V2X, and 5G. In any case, the huge volume of traffic information that CAVs produce makes issues for both the current organizations and the approaching 5G correspondence organizations. For outside network innovations, the VMBS fills in as both a client hub and an edge processing hub. For CAVs, it fills in as a base station and a data caching hub, melding correspondence and calculation. People offer both the VMBS-empowered handset and figuring for CAVs as well as the VMBS-helped wireless innovation for other wireless gadgets to achieve this. It is underlined and addressed that there are a number of research obstacles and open questions. Last but not least, the results of the simulation show that the planned VMBS-CCNA may significantly enhance throughput, latency, and the average amount of links.

  • Analysis of Sentiments on Russia-Ukraine Conflict using Machine Learning Techniques

    by Sakshi Negi, Ishaan Dawar, Rishabh Prakash, Prashant Tiwari, Megha Bhushan, Ajay Kumar

    Nowadays, the internet has become an essential part of the current scenario. Due to the rapid expansion of the social network industry, people are using its platforms to share their opinions on current events. It is important to collect and evaluate consumer feedback on products, services, and other topics. It is of utmost importance to determine if an endorsement or criticism of war operations leads to the universal message as the crisis between Russia and Ukraine has worsened and more foreign powers are involved. The abundant writings on the critical connection between public opinion to international relations support the above-mentioned objective. This study focuses mainly on Twitter text data extracted by Twitter API during the Ukraine-Russia war timeline and is available on Kaggle. Sentimental analysis using various supervised machine learning techniques, including Random Forest, Xtreme Gradient Boosting, Support Vector Machine, and Logistic Regression. Also, the comparison of their performance based on various metrics, including the accuracy and precision of the data is presented. As a result, it investigates the opinions and reactions of the global world to Russia’s invasion of Ukraine.

  • Classification of Breast Cancer using User-Defined Weighted Ensemble Voting Scheme

    by Ajay Kumar, Rama Sushil, Arvind Kumar Tiwari

    When weak classifiers, i.e., estimators, are not justifying the classification of breast cancer then ensemble learning is a way to improve the classification of cancer. The ensemble is basically an aggregator where all weak classifiers are merged to get a strong classifier. The ensemble is based on a majority voting scheme. A hard voting scheme is used to take a major vote of each classifier whereas a soft voting scheme takes the weights of the probability of each classifier. A custom based weights are assigned in this paper and the final classification of cancer using ensemble classifier is outperformed than each estimator. The highest accuracy from the proposed ensemble classifier is achieved up to 96.47% where the lowest estimator got 93.18 %. The AUC score of ensemble classifier achieved is 0.9948 which is one of the highest among all other estimators.


  • Open Access

    Machine Learning Based Approaches for Cancer Prediction: A Survey

    by Ajay Kumar, Rama Sushil, Arvind Kumar Tiwari

    Cancer is a critical disease from many years. This leads to death if it is not diagnosed at early stage. Computer Science & Engineering is used in Bioinformatics and Biomedical to diagnose and prognoses disease Cancer. This can be further directed to a field called Machine Learning where various techniques are available to predict the cancer on the basis of collected standard data sets. The datasets may have been recorded by few repositories in the world. Only we need to apply some classifiers of Machine Learning Techniques to signify the cancer in a human. In this paper, we have surveyed the research papers to compare the accuracy of different algorithm of Machine Learning about cancer depend on the given data sets and their attributes. Several papers use very common classifier technique viz. Support Vector Machine (SVM), Random Forest (RF), Naïve Bayes (NB), Decision Tree (DT), K-Nearest Neighbor (KNN), Artificial Neural Network (ANN), Fuzzy Neural Network (FNN), Radial Basis Function Network (RBFN), Shuffled Frog Leaping with Levy Flight, Particle Swarm Optimization, Back Propagation Neural Network, Multilayered Perceptron, SVM Recursive Feature Elimination etc. In order to predict cancer disease based on the given dataset, the best result among all machine learning techniques found here is SVM.


  • Open Access

    Cancer Survival Analysis Using Machine Learning

    by Kumar Ajay, Rama Sushil, Arvind Tiwari

    To improve the quality of life and increase the survival rate of individuals with cancer, early treatment and detection plays a crucial role. Early detection and diagnosis pose almost 100 percent survival rate, especially before or during stage-I. However, situations, where the detection comes during stage-IV, the rate of survival is as low as 30 percent. The quest to foster early detection has paved the way for the evolution of machine learning techniques and have emerged in response to cancer’s big data. This paper has examined Machine Learning (ML) predictive models that have been applied in the early detection of cancer, providing some of the benefits and drawbacks with which they are associated. From the results documented in most of the current literature, the ML techniques pose remarkable improvements in the prediction and classification accuracy of cancer. The implication is that in future, healthcare systems ought to combine various ML techniques with multidimensional heterogeneous data to produce more accurate results regarding cancer prediction and classification.