Paper Information | SBC BOOKS

Fill in your paper information

Language (*)

Title (*)

Keywords (*)

Abstract (*)

Neste artigo apresentamos a Keras-Prov, uma extensão à biblioteca de aprendizado profundo Keras para prover dados de proveniência. A Keras-Prov captura, armazena e gerencia metadados e dados de proveniência de experimentos de aprendizado de máquina (ML), em especial de aprendizado profundo. A Keras-Prov identifica automaticamente as transformações de dados mais comuns, como, treinamento, teste, e adaptação, para capturar os dados de proveniência. A Keras-Prov flexibiliza a captura automática, permitindo que novos dados de proveniência sejam definidos, como valores adicionais de hiperparâmetros. À gerência de proveniência por meio do SGBD colunar MonetDB, Keras-Prov adiciona uma interface de monitoramento visual e um gerador de SQL para consultas analíticas aos dados durante a evolução do treinamento e a escolha de modelos. A análise de dados da Keras-Prov, durante o treinamento, subsidia decisões de sintonia fina de hiperparâmetros. A base de dados segue a recomendação W3C PROV, favorecendo a comparação, explicação e reprodução de tais experimentos de ML. A Keras-Prov é uma solução de código aberto e pode ser obtida em https://github.com/dbpina/keras-prov.

Pages (*)

File Link

English Information

Title

Keywords

Abstract

In this paper, we introduce Keras-Prov, an extension to the Keras deep learning library to provide provenance data. Keras-Prov captures, stores, and manages metadata and data from machine learning (ML) experiments, in particular deep learning. Keras-Prov automatically identifies the most common data transformations, such as training, testing, and adapting, to capture the provenance data. Keras-Prov makes automatic capture flexible, allowing new provenance data to be defined, such as additional hyperparameter values. To the provenance management through the columnar DBMS MonetDB, Keras-Prov adds a visual monitoring interface and a SQL generator for analytical queries to the data during the training evolution and the choice of models. Keras-Prov data analysis, during training, supports hyperparameter fine-tuning decisions. The database follows the W3C PROV recommendation, favoring the comparison, explanation, and reproduction of such ML experiments. Keras-Prov is an open-source solution and can be obtained from https://github.com/dbpina/keras-prov.

(*) To change the order drag the item to the new position.

Authors

#	Name
1	Débora Pina(dbpina@cos.ufrj.br)
2	Liliane Neves(lneves@cos.ufrj.br)
3	Daniel de Oliveira(danielcmo@ic.uff.br)
4	Marta Mattoso(marta@cos.ufrj.br)

(*) To change the order drag the item to the new position.

Reference

#	Reference
1	Almeida, R. F., da Silva, W. M. C., Castro, K., de Araújo, A. P. F., Walter, M. E. T., Lifschitz, S., and Holanda, M. (2019). Managing data provenance for bioinformatics workflows using aprovbio. Int. J. Comput. Biol. Drug Des., 12(2):153–170.
2	Fairweather, E., Wittner, R., Chapman, M., Holub, P., and Curcin, V. (2020). Non-repudiable provenance for clinical decision support systems. CoRR, abs/2006.11233.
3	Fekete, J., Freire, J., and Rhyne, T. (2020). Exploring reproducibility in visualization. IEEE Computer Graphics and Applications, 40(5):108–119.
4	Freire, J., Koop, D., Santos, E., and Silva, C. T. (2008). Provenance for computational tasks: A survey. Computing in Science & Engineering, 10(3):11–21.
5	Goodfellow, I., Bengio, Y., Courville, A., and Bengio, Y. (2016). Deep learning, volume 1. MIT press Cambridge.
6	Krizhevsky, A., Sutskever, I., and Hinton, G. E. (2012). Imagenet classification with deep convolutional neural networks. In Advances in neural information processing systems, pages 1097–1105.
7	Moreau, L. and Groth, P. (2013). Provenance: an introduction to prov. Synthesis Lectures on the Semantic Web: Theory and Technology, 3(4):1–129.
8	Orr, G. B. and Muller, K.-R. (2003). Neural networks: tricks of the trade. Springer.
9	Pimentel, J. F., Murta, L., Braganholo, V., and Freire, J. (2017). noworkflow: a tool for collecting, analyzing, and managing provenance from python scripts. VLDB, 10(12).
10	Pina, D., Kunstmann, L., de Oliveira, D., Valduriez, P., and Mattoso, M. (2021). Provenance supporting hyperparameter analysis in deep neural networks. In IPAW, pages 20–38.
11	Raissi, M., Perdikaris, P., and Karniadakis, G. E. (2017). Physics informed deep learning (part i): Data-driven solutions of nonlinear partial differential equations. arXiv preprint arXiv:1711.10561.
12	Russell, S. J. and Norvig, P. (2020). Artificial Intelligence: A Modern Approach (4th Edition). Pearson.
13	Silva, V., de Oliveira, D., Valduriez, P., and Mattoso, M. (2018). Dfanalyzer: runtime dataflow analysis of scientific applications using provenance. VLDB, 11:2082–2085.
14	Souza, R., Azevedo, L., Lourenço, V., Soares, E., Thiago, R., Brandão, R., Civitarese, D., Brazil, E. V., Moreno, M., Valduriez, P., Mattoso, M., Cerqueira, R., and Netto, M. A. S. (2019). Provenance data in the machine learning lifecycle in computational science and engineering. In WORKS, pages 1–10. IEEE.

Paper Registration

Fill in your paper information

English Information

Authors

Reference