Predição de Métricas em Grafos Temporais Utilizando Redes Neurais

• Daiane M. Pereira Universidade Federal do Rio de Janeiro
• Rodrigo S. Couto Universidade Federal do Rio de Janeiro

DOI: https://doi.org/10.5753/wperformance.2021.15725

Resumo

Métricas básicas calculadas para grafos temporais, como o grau médio e o coeficiente de clusterização, são utilizadas em vários domínios, como na caracterização do fluxo de tráfego em sistemas de transporte. Apesar da aplicabilidade dessas métricas, a literatura não aborda técnicas para prever a sua evolução temporal. Para preencher essa lacuna, este trabalho aborda o uso de modelos de rede neural para prever métricas de grafos temporais. Assim, analisa-se o desempenho de uma perceptron multicamadas, de uma rede neural recorrente e de uma rede neural convolucional. Esses modelos são comparados com um modelo de base simples, mostrando desempenho satisfatório, principalmente para as redes neurais recorrentes e convolucionais.

Palavras-chave: Grafo Temporal, Rede Neural, RNN, CNN, MLP

Referências

Bergstra, J. e Bengio, Y. (2012). Random search for hyper-parameter optimization. Journal of machine learning research, 13(2).

Binkowski, M., Marti, G. e Donnat, P. (2018). Autoregressive convolutional neural networks for asynchronous time series. Em International Conference on Machine Learning, p. 580–589. PMLR.

Borovykh, A., Bohte, S. e Oosterlee, C. W. (2017). Conditional time series forecasting with convolutional neural networks. arXiv preprint arXiv:1703.04691.

Cem, E. e Sarac, K. (2016). Estimation of structural properties of online social networks at the extreme. Computer Networks, 108:323–344.

Charakopoulos, A., Karakasidis, T., Papanicolaou, P. e Liakopoulos, A. (2014). The application of complex network time series analysis in turbulent heated jets. Chaos: An Interdisciplinary Journal of Nonlinear Science, 24(2):024408.

Cui, Z., Henrickson, K., Ke, R. e Wang, Y. (2019). Traffic graph convolutional recurrent neural network: A deep learning framework for network-scale traffic learning and forecasting. IEEE Transactions on Intelligent Transportation Systems, 21(11):4883–4894.

De Gooijer, J. G. e Hyndman, R. J. (2006). 25 years of time series forecasting. International journal of forecasting, 22(3):443–473.

Frank, R. J., Davey, N. e Hunt, S. P. (2001). Time series prediction and neural networks. Journal of intelligent and robotic systems, 31(1):91–103.

Hegeman, T. e Iosup, A. (2018). Survey of graph analysis applications. arXiv preprint arXiv:1807.00382.

Hewamalage, H., Bergmeir, C. e Bandara, K. (2020). Recurrent neural networks for time series forecasting: Current status and future directions. International Journal of Forecasting.

Hua, Y., Zhao, Z., Li, R., Chen, X., Liu, Z. e Zhang, H. (2019). Deep learning with long short-term memory for time series prediction. IEEE Communications Magazine, 57(6):114–119.

Hyndman, R. J. e Athanasopoulos, G. (2018). Forecasting: principles and practice. OTexts.

Kim, S. e Kang, M. (2019). Financial series prediction using attention lstm. arXiv preprint arXiv:1902.10877.

Li, Y., Yu, R., Shahabi, C. e Liu, Y. (2017). Diffusion convolutional recurrent neural network: Data-driven traffic forecasting. arXiv preprint arXiv:1707.01926.

Lim, B. e Zohren, S. (2021). Time-series forecasting with deep learning: a survey. Philosophical Transactions of the Royal Society A, 379(2194):20200209.

Makridakis, S., Spiliotis, E. e Assimakopoulos, V. (2018). Statistical and machine learning forecasting methods: Concerns and ways forward. PloS one, 13(3):e0194889.

Mbah, T. J., Ye, H., Zhang, J. e Long, M. (2021). Using lstm and arima to simulate and predict limestone price variations. Mining, Metallurgy & Exploration, p. 1–14.

Nakamoto, S. (2019). Bitcoin: A peer-to-peer electronic cash system. Relatório técnico, Manubot.

Nicosia, V., Tang, J., Mascolo, C., Musolesi, M., Russo, G. e Latora, V. (2013). Graph metrics for temporal networks. Em Temporal networks, p. 15–40. Springer.

Nielsen, A. (2019). Practical time series analysis: prediction with statistics and machine learning. ”O’Reilly Media, Inc.”.

Santoro, N., Quattrociocchi, W., Flocchini, P., Casteigts, A. e Amblard, F. (2011). Timevarying graphs and social network analysis: Temporal indicators and metrics. arXiv preprint arXiv:1102.0629.

Selvin, S., Vinayakumar, R., Gopalakrishnan, E., Menon, V. K. e Soman, K. (2017). Stock price prediction using lstm, rnn and cnn-sliding window model. Em 2017 international conference on advances in computing, communications and informatics (icacci), p. 1643–1647. IEEE.

Siami-Namini, S., Tavakoli, N. e Namin, A. S. (2018). A comparison of arima and lstm in forecasting time series. Em 2018 17th IEEE International Conference on Machine Learning and Applications (ICMLA), p. 1394–1401. IEEE.

Siami-Namini, S., Tavakoli, N. e Namin, A. S. (2019). A comparative analysis of forecasting financial time series using arima, lstm, and bilstm. arXiv preprint ar-Xiv:1911.09512.

Tang, J., Liu, F., Zhang,W., Zhang, S. eWang, Y. (2016). Exploring dynamic property of traffic flow time series in multi-states based on complex networks: Phase space reconstruction versus visibility graph. Physica A: Statistical Mechanics and its Applications, 450:635–648.

Tang, J., Wang, Y., Wang, H., Zhang, S. e Liu, F. (2014). Dynamic analysis of traffic time series at different temporal scales: A complex networks approach. Physica A: Statistical Mechanics and its Applications, 405:303–315.

Yamak, P. T., Yujian, L. e Gadosey, P. K. (2019). A comparison between arima, lstm, and gru for time series forecasting. Em Proceedings of the 2019 2nd International Conference on Algorithms, Computing and Artificial Intelligence, p. 49–55.

Yan, S. (2015). Understanding lstm networks. Online, acessado em Janeiro de 2021, 11.

Yu, B., Yin, H. e Zhu, Z. (2017). Spatio-temporal graph convolutional networks: A deep learning framework for traffic forecasting. arXiv preprint arXiv:1709.04875.