Treinamento Paralelo de Redes Neurais usando o Modelo de Atores
Resumo
O Modelo de Atores usa concorrência baseada em processos que não compartilham memória e que se comunicam por troca de mensagens, e tem sido usado no treinamento federado de redes neurais. Esse artigo é um trabalho em andamento que investiga o treinamento paralelo de redes neurais em dispositivos multiprocessados, utilizando o Modelo de Atores para distribuir e processar batches de treinamento, com objetivo de melhorar algoritmos de treinamento.Referências
(2023). Matrex. WWW page, https://hex.pm/packages/matrex.
(2023). The Elixir Language. WWW page, https://elixir-lang.org/.
(2023). The Erlang Language. WWW page, https://www.erlang.org/.
(2023). The MNIST Database. WWW page, http://yann.lecun.com/exdb/mnist/.
(2023). The Scala Language. WWW page, https://www.scala-lang.org/.
Hewitt, C., Bishop, P., and Steiger, R. (1973). Session 8 formalisms for artificial intelligence a universal modular actor formalism for artificial intelligence. In Advance Papers of the Conference, volume 3, page 235. Stanford Research Institute Menlo Park, CA.
Monteiro, A., De Oliveira, M., De Oliveira, R., and Da Silva, T. (2018). Embedded application of convolutional neural networks on raspberry pi for shm. Electronics Letters, 54(11):680-682.
Ruokolainen, T. et al. (2017). Development of a distributed web server utilizing elixir.
Srirama, S. N., Dick, F. M. S., and Adhikari, M. (2021). Akka framework based on the actor model for executing distributed fog computing applications. Future Generation Computer Systems, 117:439-452.
Srirama, S. N. and Vemuri, D. (2023). Canto: An actor model-based distributed fog framework supporting neural networks training in iot applications. Computer Communications, 199:1-9.
Xu, R., Michala, A. L., and Trinder, P. (2022). Caefl: composable and environment aware federated learning models. In Proceedings of the 21st ACM SIGPLAN International Workshop on Erlang, pages 9-20.
Yuan, B., Wolfe, C. R., Dun, C., Tang, Y., Kyrillidis, A., and Jermaine, C. (2022). Distributed learning of fully connected neural networks using independent subnet training. Proc. VLDB Endow., 15(8):1581-1590.
(2023). The Elixir Language. WWW page, https://elixir-lang.org/.
(2023). The Erlang Language. WWW page, https://www.erlang.org/.
(2023). The MNIST Database. WWW page, http://yann.lecun.com/exdb/mnist/.
(2023). The Scala Language. WWW page, https://www.scala-lang.org/.
Hewitt, C., Bishop, P., and Steiger, R. (1973). Session 8 formalisms for artificial intelligence a universal modular actor formalism for artificial intelligence. In Advance Papers of the Conference, volume 3, page 235. Stanford Research Institute Menlo Park, CA.
Monteiro, A., De Oliveira, M., De Oliveira, R., and Da Silva, T. (2018). Embedded application of convolutional neural networks on raspberry pi for shm. Electronics Letters, 54(11):680-682.
Ruokolainen, T. et al. (2017). Development of a distributed web server utilizing elixir.
Srirama, S. N., Dick, F. M. S., and Adhikari, M. (2021). Akka framework based on the actor model for executing distributed fog computing applications. Future Generation Computer Systems, 117:439-452.
Srirama, S. N. and Vemuri, D. (2023). Canto: An actor model-based distributed fog framework supporting neural networks training in iot applications. Computer Communications, 199:1-9.
Xu, R., Michala, A. L., and Trinder, P. (2022). Caefl: composable and environment aware federated learning models. In Proceedings of the 21st ACM SIGPLAN International Workshop on Erlang, pages 9-20.
Yuan, B., Wolfe, C. R., Dun, C., Tang, Y., Kyrillidis, A., and Jermaine, C. (2022). Distributed learning of fully connected neural networks using independent subnet training. Proc. VLDB Endow., 15(8):1581-1590.
Publicado
10/05/2023
Como Citar
BARTSCH, Ariam Moresco; DU BOIS, Andre Rauber.
Treinamento Paralelo de Redes Neurais usando o Modelo de Atores. In: ESCOLA REGIONAL DE ALTO DESEMPENHO DA REGIÃO SUL (ERAD-RS), 23. , 2023, Porto Alegre/RS.
Anais [...].
Porto Alegre: Sociedade Brasileira de Computação,
2023
.
p. 81-84.
ISSN 2595-4164.
DOI: https://doi.org/10.5753/eradrs.2023.230096.
