Automated machine learning: learning to learn

• Grant number: 18/14819-5
• Support Opportunities: Scholarships in Brazil - Doctorate (Direct)
• Start date: December 01, 2018
• Status: Discontinued
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Computing Methodologies and Techniques
• Principal Investigator: André Carlos Ponce de Leon Ferreira de Carvalho
• Grantee: Edesio Pinto de Souza Alcobaça Neto
• Host Institution: Instituto de Ciências Matemáticas e de Computação (ICMC). Universidade de São Paulo (USP). São Carlos , SP, Brazil
• Associated research grant: 13/07375-0 - CeMEAI - Center for Mathematical Sciences Applied to Industry, AP.CEPID

Abstract

A large amount of data is currently generated by means of sensors, smartphones, and the internet. However, analyzing it manually is not an easy task and the use of computational techniques to extract useful knowledge is handy. The Data Science and machine learning fields study such techniques to induce models from datasets. However, inducing a good model from raw data is not trivial. It requires several steps, such as preprocessing, algorithm selection and hyperparameters configuration. These steps have high computational cost and they require decision-making, often difficult for lay users. Even for experts, such tasks are usually manual and repetitive, and the time consumed could be used in more critical activities. To address those difficulties, this project will research and develop an automated machine learning system. Such systems automate different tasks related to the application of machine learning algorithms on datasets, considering also the data preprocessing and the post-processing of the results. This research will approach: preprocessing (cleaning, feature selection), modeling (appropriate choice of algorithms and hyperparameters adjustment) and post-processing (model evaluation and user report). The main focus will be preprocessing, which includes analyzing how it influences the performance of machine learning algorithms in an automated machine learning system. In addition, the use of meta-learning in the automated learning system will be investigated. We hope the system to assist both expert and lay users in the tasks required for data analysis, letting them focus on problem analysis and results by avoiding time-consuming manual adjustments. (AU)

Scientific publications

(6)

(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)

CASSAR, R. DANIEL; MASTELINI, SAULO MARTIELLO; BOTARI, TIAGO; ALCOBACA, EDESIO; DE CARVALHO, C. P. L. F. ANDRE; ZANOTTO, D. EDGAR. Predicting and interpreting oxide glass properties by machine learning using large datasets. CERAMICS INTERNATIONAL, v. 47, n. 17, p. 23958-23972, SEP 1 2021. (18/14819-5, 13/07375-0, 17/12491-0, 13/07793-6, 18/07319-6)

MANTOVANI, RAFAEL G.; ROSSI, ANDRE L. D.; ALCOBACA, EDESIO; VANSCHOREN, JOAQUIN; DE CARVALHO, ANDRE C. P. L. F.. A meta-learning recommender system for hyperparameter tuning: Predicting when tuning improves SVM classifiers. INFORMATION SCIENCES, v. 501, p. 193-221, OCT 2019. (12/23114-9, 15/03986-0, 18/14819-5)

ALCOBACA, EDESIO; MASTELINI, SAULO MARTIELLO; BOTARI, TIAGO; PIMENTEL, BRUNO ALMEIDA; CASSAR, DANIEL ROBERTO; DE LEON FERREIRA DE CARVALHO, ANDRE CARLOS PONCE; ZANOTTO, EDGAR DUTRA. Explainable Machine Learning Algorithms For Predicting Glass Transition Temperatures. ACTA MATERIALIA, v. 188, p. 92-100, APR 15 2020. (17/12491-0, 13/07375-0, 18/07319-6, 17/06161-7, 17/20265-0, 13/07793-6, 18/14819-5)

GARCIA, LUIS P. F.; RIVOLLI, ADRIANO; ALCOBACA, EDESIO; LORENA, ANA C.; DE CARVALHO, ANDRE C. P. L. F.. Boosting meta-learning with simulated data complexity measures. Intelligent Data Analysis, v. 24, n. 5, p. 1011-1028, 2020. (12/22608-8, 13/07375-0, 18/14819-5, 16/18615-0)

MASTELINI, SAULO MARTIELLO; CASSAR, DANIEL R.; ALCOBACA, EDESIO; BOTARI, TIAGO; DE CARVALHO, ANDRE C. P. L. F.; ZANOTTO, EDGAR D.. Machine learning unveils composition-property relationships in chalcogenide glasses. ACTA MATERIALIA, v. 240, p. 13-pg., 2022-09-09. (18/14819-5, 13/07793-6, 17/12491-0, 18/07319-6, 17/06161-7, 13/07375-0)

ALCOBACA, EDESIO; SIQUEIRA, FELIPE; RIVOLLI, ADRIANO; GARCIA, LUIS P. F.; OLIVA, JEFFERSON T.; DE CARVALHO, ANDRE C. P. L. F.. MFE: Towards reproducible meta-feature extraction. JOURNAL OF MACHINE LEARNING RESEARCH, v. 21, 2020. (13/07375-0, 18/14819-5, 16/18615-0)