Reinforcement Learning to Manage Energy Efficient Supply Chains A Kannagi, Savita, Pankaj Kumar Goswami 2024 International Conference on Optimization Computing and Wireless Communication Icocwc 2024, 2024 Reinforcement getting to know (RL) has emerged as an ability method to deal with electricity green supply chains and improve the sustainability of operations. This paper critiques the software of RL in supply chain management, exploring the primary programs, methodologies, and approaches of incorporating RL in power structures. We overview recent advances in RL that would be implemented to supply chain power modeling, in addition to the benefits and demanding situations that can stand up from using RL for superior management of delivery chains. Further, we provide a conclusion on the blessings of RL as a tool for managing power green supply chains and advise capacity programs for research that explores how RL can be used to enhance the sustainability of operations.
Assessing Optimal Hyper parameters of Deep Neural Networks on Cancers Datasets Pankaj Kumar Goswami, A Kannagi, Anubhav Sony 2024 International Conference on Optimization Computing and Wireless Communication Icocwc 2024, 2024 This paper studies the most valuable hyperparameters of deep neural networks applied to most cancer datasets. We appoint a mixture of looking algorithms, including grid seeks, random search, and Bayesian optimization, to discover the best mixture of hyperparameters for deep neural networks. The overall performance of the one-of-a-kind algorithms is evaluated towards present most cancers datasets and as compared towards each other. Outcomes show that Bayesian optimization becomes the maximum green and correct technique for finding the most fulfilling hyperparameter for our goal deep neural networks. This research can provide precious insight to practitioners who layout and build deep-mastering models for most cancer datasets. Furthermore, it also helps to optimize the performance of the trained neural networks while applied to this specific trouble location. The painting aims to assess the most beneficial hyperparameters of deep neural networks (DNNs) on most cancer datasets. DNNs are increasingly employed within the class and analysis of cancer datasets due to their ability to capture complicated styles and hit upon relationships between relevant capabilities. However, the effectiveness of those models is somewhat affected by the layout and selection of hyperparameters, which govern their education and represent a critical factor in the model optimization manner. In this painting, we optimize the choice of hyperparameters for a DNN using a grid search approach for every dataset, one after the other. Primarily, we optimize several parameters, along with the number of layers, neurons in keeping with layer, activation functions, studying fee, range of epochs, batch size, and dropout charge. The performance of the optimized DNN version is then evaluated by studying its accuracy, AUROC, and precision while evaluating on a take-a-look-at the set. Consequences show that extensive improvements in overall performance may be performed while the most reliable hyperparameters are chosen.
The Smart Predictive Network Maintenance Model using AI/ML Time Series Analysis in Cloud Networks Prabha S., Ramakant Upadhyay, Pankaj Kumar Goswami 2024 International Conference on Optimization Computing and Wireless Communication Icocwc 2024, 2024 The ever-growing amount of facts generated from cloud networks requires a new predictive version for community renovation. System learning (ML) and synthetic Intelligence (AI) have been carried out to improve the accuracy of predictive fashions for community maintenance. This paper presents a new method of using time series evaluation with ML and AI to design a clever, predictive community protection model for cloud networks. The proposed version is based on finding patterns and correlations from ancient statistics using AI and ML techniques for time collection analysis. This version uses various features, including packet loss, latency, throughput, unique network popularity metrics, CPU utilization, memory, disk space, and community traffic. The model can detect anomalous behaviors inside the community in early degrees and may provide predictions of destiny network activities. The proposed version can additionally conform to converting network situations through superior self-mastering ML algorithms. Universal, the proposed model could permit higher choice-making and streamline the network maintenance tasks employing offering well-timed records and insights about the community's typical health.
Smart Wideband Sensing Antenna for Cognitive Radio Applications Sakshi Singh, Pankaj Kumar Goswami, Garima Goswami, Jitesh Verma Proceedings of the 2021 10th International Conference on System Modeling and Advancement in Research Trends Smart 2021, 2021