This article tackles the critical challenge of optimizing the buffer pool, a core component of Database Management Systems (DBMS) that caches frequently accessed data pages, where manual configuration often proves inadequate in dynamic, high-demand environments. To address this gap, we present an automated, data-driven methodology that combines advanced Machine Learning techniques with Bayesian optimization. Our approach follows a systematic three-phase process: (1) Exploratory Factor Analysis (EFA) coupled with K-means clustering to uncover latent factors and reduce the dimensionality of performance metrics; (2) LASSO regression to identify and rank the most influential configuration parameters; and (3) Bayesian optimization using Gaussian Process modeling with acquisition functions (Expected Improvement, Probability of Improvement, and Upper Confidence Bound) to fine-tune buffer pool settings. The main contributions of this work include a novel automated framework for DBMS tuning that simplifies configuration, enhances memory management, and boosts performance efficiency. We validated the proposed solution using real workloads collected from a large-scale financial system in Latin America, achieving up to a 45% reduction in maximum data access wait times, confirming improvements in performance and scalability.

Research in Natural Language Processing, bio-medicine, and computer vision achieved excellent results in machine learning due to the success of the Transformer-based models. However, these excellent results depend on the labeled high-quality and large-scale datasets. If one of these requirements is not met, the model may lack generalization ability, and its performance will be unsatisfactory. To address these issues, this research proposes a Generic Multimodal Gradient-Based Meta Framework (GeMGF) trained from scratch to avoid language bias, learns from a few data, and reduces the model degradation trained on a finite dataset. GeMGF was evaluated using the benchmark dataset CUB-200-2011 for the text and image classification tasks. The results show that GeMGF outperforms the state-of-the-art models with 93.2% accuracy. GeMGF is simple, efficient, and adaptable to other data modalities and fields.