Adaptive filtering and machine learning: applications in diffusion networks, soft sensors and classification of cardiac arrhythmias

• Grant number: 21/02063-6
• Support Opportunities: Regular Research Grants
• Start date: August 01, 2021
• End date: July 31, 2023
• Field of knowledge: Engineering - Electrical Engineering - Telecommunications
• Principal Investigator: Magno Teófilo Madeira da Silva
• Grantee: Magno Teófilo Madeira da Silva
• Host Institution: Escola Politécnica (EP). Universidade de São Paulo (USP). São Paulo , SP, Brazil
• City of the host institution: São Paulo
• Associated researchers: Renato Candido

Abstract

Distributed signal processing has attracted attention in the scientific community due to its several advantages over centralized approaches. In this field, sampling and censoring techniques have been topics of intense research, since the cost associated with measuring, processing and/or transmitting data throughout the entire network may be prohibitive in certain situations. The aim of this project is to deeply study sampling and censoring techniques, by proposing improvements to the algorithms that were recently proposed in the literature. These algorithms will be extended to different diffusion networks, such as multitask, kernel-based, and asynchronous networks. Distributed versions of kernel adaptive filters will also be addressed and compared to distributed neural networks for thesolution of nonlinear problems in diffusion networks. The Gram-Schmidt-based sparsification technique for dictionary will be also applied to kernel principal component analysis. In the machine learning field, we intend to use soft sensors for fault detection and diagnosisin air conditioning systems, enabling corrections before the increase in the consumption of electrical energy. Finally, we indent to use neural networks to classify cardiac arrhythmias, taking into account the separation of the patients into training and testing sets. (AU)

Scientific publications

(7)

(References retrieved automatically from Web of Science and SciELO through information on FAPESP grants and their corresponding numbers as mentioned in the publications by the authors)

TIGLEA, DANIEL G.; CANDIDO, RENATO; SILVA, MAGNO T. M.. A Variable Step Size Adaptive Algorithm With Simple Parameter Selection. IEEE SIGNAL PROCESSING LETTERS, v. 29, p. 5-pg., 2022-01-01. (21/02063-6)

TIGLEA, DANIEL GILIO; CANDIDO, RENATO; SILVA, MAGNO T. M.. Adaptive diffusion networks: An overview. Signal Processing, v. 223, p. 27-pg., 2024-06-15. (21/02063-6)

TIGLEA, DANIEL G.; CANDIDO, RENATO; SILVA, MAGNO T. M.. On the Impact of Random Node Sampling on Adaptive Diffusion Networks. IEEE TRANSACTIONS ON SIGNAL PROCESSING, v. 72, p. 17-pg., 2024-01-01. (21/02063-6)

TIGLEA, DANIEL G.; CANDIDO, RENATO; SILVA, MAGNO T. M.. Can Adaptive Diffusion Networks Do Better with Less Data?. 2024 19TH INTERNATIONAL SYMPOSIUM ON WIRELESS COMMUNICATION SYSTEMS, ISWCS 2024, v. N/A, p. 6-pg., 2024-01-01. (21/02063-6)

BORGES, VINICIUS S.; SILVA, MAGNO T. M.; EISENCRAFT, MARCIO. Chaotic properties of an FIR filtered Hénon map. COMMUNICATIONS IN NONLINEAR SCIENCE AND NUMERICAL SIMULATION, v. 131, p. 13-pg., 2024-01-23. (21/02063-6)

TIGLEA, DANIEL G.; CANDIDO, RENATO; SILVA, MAGNO T. M.. Bias-Variance Decomposition of the Mean-Square Deviation of the LMS Algorithm: Transient and Steady-State Analysis. CIRCUITS SYSTEMS AND SIGNAL PROCESSING, v. N/A, p. 32-pg., 2025-02-09. (21/02063-6)

TIGLEA, DANIEL GILIO; CANDIDO, RENATO; SILVA, MAGNO T. M.. An adaptive algorithm for sampling over diffusion networks with dynamic parameter tuning and change detection mechanisms. DIGITAL SIGNAL PROCESSING, v. 127, p. 18-pg., 2022-05-16. (21/02063-6)