Ariel Pincay
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Ariel Pincay has a total of 8 co-authors. Here are the top co-authors by number of shared publications:
- Francisco Hidrobo
Yachay Tech University
2 shared publications - Carlos David Castro-Rodriguez
Communication Networks and Intelligent Services Research Group (ComNet Innova YT), School of Mathematical and Computational Sciences, Yachay Tech University, Hacienda San José s/n y Proyecto Yachay, Urcuquí 100119, Ecuador
1 shared publication - Cristofer Motoche-Monar
Phage Therapy Group, Yachay Tech University, Urcuqui 100115, Ecuador
1 shared publication - Freddy Valenzuela
Communication Networks and Intelligent Services Research Group (ComNet Innova YT), School of Mathematical and Computational Sciences, Yachay Tech University, Hacienda San José s/n y Proyecto Yachay, Urcuquí 100119, Ecuador
1 shared publication - Joseph L. Tipan-Vera
Communication Networks and Intelligent Services Research Group (ComNet Innova YT), School of Mathematical and Computational Sciences, Yachay Tech University, Hacienda San José s/n y Proyecto Yachay, Urcuquí 100119, Ecuador
1 shared publication - José Antonio Castillo
Phage Therapy Group, Yachay Tech University, Urcuqui 100115, Ecuador
1 shared publication
Latest Publications
- Open Access
- 18 September 2026
- Published by MDPI AG in Engineering Proceedings
- Vol. 153 (1) , 3
Comparative Study of Federated Learning Architectures on Resource-Constrained Edge Devices
This paper presents a comparative experimental evaluation of centralized, semi-decentralized, and hierarchical federated learning (FL) architectures combined with FedAvg, FedProx, and FedNova for premature non-communicable disease (NCD) mortality prediction. Using the Ecuadorian mortality registry (89,877 records), we construct a non-IID institutional setting with quantity, label, and feature skew across five hospitals and deploy all configurations on a six-device Raspberry Pi 4 cluster. The evaluation jointly considers predictive performance, convergence, communication overhead, energy estimates, CPU usage, RAM usage, temperature, and straggler delay. Hierarchical FedNova achieves the highest global F1-score (0.808) and accuracy (0.941), with near-zero straggler delay and the lowest cumulative energy among hierarchical configurations (15.35 J). Semi-decentralized FL reduces communication and energy cost but is sensitive to lottery-based aggregator selection under hardware volatility. Centralized FedNova provides the lowest estimated cumulative energy ( J) at the cost of lower predictive performance. The findings provide practical guidance for FL deployment in resource-limited, data-heterogeneous public-health environments under the evaluated hardware and data-partition conditions.- 18 September 2026
- Published by MDPI AG in Engineering Proceedings
- Vol. 153 (1) , 3
Open AccessComparative Study of Federated Learning Architectures on Resource-Constrained Edge Devices
This paper presents a comparative experimental evaluation of centralized, semi-decentralized, and hierarchical federated learning (FL) architectures combined with FedAvg, FedProx, and FedNova for premature non-communicable disease (NCD) mortality prediction. Using the Ecuadorian mortality registry (89,877 records), we construct a non-IID institutional setting with quantity, label, and feature skew across five hospitals and deploy all configurations on a six-device Raspberry Pi 4 cluster. The evaluation jointly considers predictive performance, convergence, communication overhead, energy estimates, CPU usage, RAM usage, temperature, and straggler delay. Hierarchical FedNova achieves the highest global F1-score (0.808) and accuracy (0.941), with near-zero straggler delay and the lowest cumulative energy among hierarchical configurations (15.35 J). Semi-decentralized FL reduces communication and energy cost but is sensitive to lottery-based aggregator selection under hardware volatility. Centralized FedNova provides the lowest estimated cumulative energy ( J) at the cost of lower predictive performance. The findings provide practical guidance for FL deployment in resource-limited, data-heterogeneous public-health environments under the evaluated hardware and data-partition conditions. - Open Access
Computational Protein Redesign of Bacteriophages by Using Evolutionary Algorithms
Protein–protein interactions are a fundamental component of most biological processes in living organisms. Therefore, enhancing these molecular interactions is of major importance in the life sciences field. In this paper we present the redesign of a protein–protein complex using single-elitist and multi-objective evolutionary algorithms. We focus on a structural complex composed of a bacteriophage protein interacting with a bacterial protein, emphasizing the interaction zone, which is defined by the closest distance between residues from proteins and comprises thirty-eight positions. Our approach fuses physics-based energy calculations with data-driven models to maximize the effectiveness of the search process. By simultaneously minimizing interaction energy and maximizing the log-likelihood ratio, the proposed algorithms show a balance between thermodynamic stability and biological sequence plausibility. This hybrid strategy guides the mutation of specific residues, enabling the identification of optimal solutions that are both physically robust and evolutionarily relevant. Results demonstrate that the evolved Pareto optimal set exhibits a significant improvement in binding affinity, with mean interface energy decreasing from +32 to −60 REU (Rosetta Energy Units). Furthermore, the analysis identifies key conserved residue positions, validating the capability of the framework to produce energetically favorable and biologically consistent protein designs.
Open AccessComputational Protein Redesign of Bacteriophages by Using Evolutionary Algorithms
Protein–protein interactions are a fundamental component of most biological processes in living organisms. Therefore, enhancing these molecular interactions is of major importance in the life sciences field. In this paper we present the redesign of a protein–protein complex using single-elitist and multi-objective evolutionary algorithms. We focus on a structural complex composed of a bacteriophage protein interacting with a bacterial protein, emphasizing the interaction zone, which is defined by the closest distance between residues from proteins and comprises thirty-eight positions. Our approach fuses physics-based energy calculations with data-driven models to maximize the effectiveness of the search process. By simultaneously minimizing interaction energy and maximizing the log-likelihood ratio, the proposed algorithms show a balance between thermodynamic stability and biological sequence plausibility. This hybrid strategy guides the mutation of specific residues, enabling the identification of optimal solutions that are both physically robust and evolutionarily relevant. Results demonstrate that the evolved Pareto optimal set exhibits a significant improvement in binding affinity, with mean interface energy decreasing from +32 to −60 REU (Rosetta Energy Units). Furthermore, the analysis identifies key conserved residue positions, validating the capability of the framework to produce energetically favorable and biologically consistent protein designs.