A significant part of global quantum computing research has been conducted based on quantum mechanics, which can now be used with quantum computers. However, designing a quantum algorithm requires a deep understanding of quantum mechanics and physics procedures. This work presents a generic quantum “black box” for entropy calculation. It does not depend on the data type and can be applied to building and maintaining machine learning models. The method has two main advantages. First, it is accessible to those without preliminary knowledge of quantum computing. Second, it is based on the quantum circuit with a constant depth of three, which is equivalent to three operations the circuit would perform to achieve the same result. We implemented our method using the IBM simulator and tested it over different types of input. The results showed a high correspondence between the classical and quantum computations that raised an error of up to 8.8e−16 for different lengths and types of information.

Clustering techniques are convenient tools for preparing and organizing unstructured and unclassified data. Depending on the data, they can be used to prepare for an analysis or to gain insight. However, choosing a clustering technique can be challenging when dealing with high-dimensional datasets. Most often, application requirements and data distribution need to be considered. Since clustering is defined as a complex problem to calculate, different algorithms may produce different results that meet the application's needs. This study presents an automated threshold-based and goal-oriented clustering procedure. It is based on the AutoML mechanism to estimate the most suitable hyperparameters according to predefined needs and can learn four clustering performance metrics thresholds for a given dataset. The significant advantages of this method are the automatic selection of clustering technique (i.e., partitional, hierarchical, density-based, or graph-based) and the ability to determine the output dynamically, according to predefined goals. We tested our method over four datasets and analyzed the results according to different goals. The results show that our method improved the silhouette score by 549.5% (from 0.105 to 0.682) compared to popular and commonly used K-means. Furthermore, clustering based on multiple metrics yielded more information than clustering by a single metric.

This paper proposes a novel hybrid approach that combines unsupervised feature extraction through clustering and unsupervised feature selection for data reduction, specifically targeting high-dimensional data. The proposed method employs K-means clustering for feature extraction, where cluster membership serves as a new feature representation, capturing the inherent data characteristics. Subsequently, the K-Nearest Neighbors (KNN) and Random Forest algorithms are utilized for supervised feature selection, identifying the most relevant feature to enhance model performance. This hybrid approach leverages the strengths of both unsupervised and supervised learning techniques. The new algorithm was applied to 13 different tabular datasets, with 9 datasets showing significant improvements across various performance metrics (accuracy, precision, recall, and F1-score) in both KNN and Random Forest models, despite substantial feature reduction. In the remaining four datasets, we achieved substantial dimensionality reduction with only negligible performance decreases. This improvement in performance while reducing dimensionality highlights the potential of the proposed method within the procedure, where datasets are treated without prior knowledge or assumptions. The proposed method offers a promising solution for handling high-dimensional data, enhancing model performance while maintaining interpretability and ease of integration within the proposed frameworks, with the ability to be irrespective of supervised or unsupervised designation datasets while reducing the dependency on a target or label features.

Modern large language models have achieved remarkable growth and performance across domains, yet their intense use of resources and high computational costs present challenges to scalability and sustainability. Current attempts to surpass baseline (naïve) AutoDL (Automated Deep Learning) models often rely on complex manipulations to yield marginal accuracy gains while demanding deep domain knowledge and computational intensity. To address known inefficiencies in computation and implementation, this study proposes a strategic approach for enhancing processing without compromising model accuracy or performance through a simplified, scalable methodology. We present a novel AutoDL weight optimization model method that analyzes the most accurate deep learning starting point and achieves the highest outcomes while considering the additional “presetting” analysis overhead. Using 20 real-world datasets, we conducted experiments across three models, six weight configurations, and ten seeds, totaling 62,400 epochs. In all experiments, the optimized model outperformed baselines, achieving higher accuracy across every dataset while reducing preprocessing to only two epochs per seed. These results demonstrate that minimal preprocessing—limited to two epochs per seed—can substantially lower computational demand while maintaining precision. As the global demand for AI deployment accelerates, this conservation-oriented approach will be critical to sustaining innovation within resource and infrastructural constraints, enabling advances in computational sustainability and responsible AI development, tangible savings across multiple dimensions of resource consumption, and broader access to deep learning technologies.