Abstract
Recurrent Neural Networks (RNNs) have been widely used for sequences analysis and classification. Generally, the sequences are a set of samples following a specific order, like a time-based process or a structured dataset. This type of neural network is very efficient for exploring sequences patterns and other relevant features highlighting temporal behavior and dependencies. This is accomplished because the information loops within the different stages of the network, and in this process, it remembers and tracks features at different segments of the data. In this work, we are interested in exploring how an RNN based on Long-Short Term Memory (LSTM) units behaves in a classification problem when the dataset of sequences are organized in different order and lengths. That is, the same information is presented to the network, but the order of the samples within the sequences and the length of the sequences are different in each experiment. In order to evaluate this effect, we used five datasets of grayscale images of 28 \(\times \) 28 pixels (MNIST, MNIST-C, notMNIST, FashionMNIST, and Sign Language MNIST). For every experiment, we segmented the images in different sizes and orders and built a set of sequences consisting of vectors of pixels organized following three different rules, and on each case, we set the sequences to a specifically fixed length. The results bring to light that good accuracies can be achieved for different sequences configurations. We considered the 28 \(\times \) 28 configuration as the baseline for reference. We found that this baseline generally leads to high accuracies, but for some datasets it is not the best one. We believe that this study may be useful for video tagging and for general image description.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
References
Bai, S., Kolter, J.Z., Koltun, V.: An empirical evaluation of generic convolutional and recurrent networks for sequence modeling (2018). http://arxiv.org/abs/1803.01271. Accessed 29 June 2020
Breuel, T.M.: Benchmarking of LSTM Networks (2015). https://arxiv.org/abs/1508.02774. Accessed 01 June 2020
Bulatov, Y.: The notMNIST dataset (2011). http://yaroslavvb.blogspot.com/2011/09/notmnist-dataset.html. Accessed 03 July 2020
Chalupka, K., Perona, P., Eberhardt, F.: Fast conditional independence test for vector variables with large sample sizes (2018). http://arxiv.org/abs/1804.02747. Accessed 25 June 2020
Cielen, D., Meysman, A.D.B., Ali, M.: Introducing Data Science, 1st edn. Manning Publications Co., Shelter Island (2016)
Cover, T.M., Thomas, J.A.: Elements of Information Theory, 2nd edn. Wiley-Interscience, Hoboken (2006)
Dietterich, T.G.: Machine learning for sequential data: a review. In: Caelli, T., Amin, A., Duin, R.P.W., de Ridder, D., Kamel, M. (eds.) SSPR /SPR 2002. LNCS, vol. 2396, pp. 15–30. Springer, Heidelberg (2002). https://doi.org/10.1007/3-540-70659-3_2
Gonzales, R.G., Woods, R.E.: Digital Image Processing, 4th edn. Pearson, New York (2018)
Goodfellow, I., Bengio, Y., Courville, A.: Deep Learning, Final edn. The MIT Press, Cambridge (2016)
Graves, A.: Generating sequences with recurrent neural networks (2013). http://arxiv.org/abs/1308.0850. Accessed 17 June 2020
Kingma, D.P., Ba, J.L.: Adam: a method for stochastic optimization (2014). https://arxiv.org/abs/1412.6980. Accessed 09 July 2020
LeCun, Y.: The MNIST database of handwritten digits (1998). http://yann.lecun.com/exdb/mnist/index.html. Accessed 15 June 2020
McMillan, B., Slepian, D.: Information theory. In: Encyclopedia of Cognitive Science, pp. 1151–1157 (2006)
Mu, N., Gilmer, J.: MNIST-C: a robustness benchmark for computer vision (2019). https://zenodo.org/record/3239543#.X0UgJC2ZO00. Accessed 21 June 2020
Ng, R.: Exploring notMNIST with TensorFlow (2018). https://www.ritchieng.com/machine-learning/deep-learning/tensorflow/notmnist/. Accessed 04 July 2020
Ostmeyer, J., Cowell, L.: Machine learning on sequential data using a recurrent weighted average. Neurocomputing 331, 281–288 (2019)
Patterson, J., Gibson, A.: Deep Learning, a Practitioner’s Approach, 1st edn. O’Reilly Media, Sebastopol (2017)
Raschka, S., Mirjalili, V.: Python Machine Learning, 3rd edn. Packt Publishing Ltd., Birmingham (2019)
Runge, J.: TIGRAMITE - Causal discovery for time series datasets (2019). https://github.com/jakobrunge/tigramite#tigramite-causal-discovery-for-time-series-datasets. Accessed 17 June 2020
Runge, J., Nowack, P., Kretschmer, M., Flaxman, S., Sejdinovic, D.: Detecting and quantifying causal associations in large nonlinear time series datasets. Sci. Adv. 5(11), eaau4996 (2019)
Schak, M., Gepperth, A.: A study on catastrophic forgetting in deep LSTM networks. In: Tetko, I.V., Kůrková, V., Karpov, P., Theis, F. (eds.) ICANN 2019. LNCS, vol. 11728, pp. 714–728. Springer, Cham (2019). https://doi.org/10.1007/978-3-030-30484-3_56
Schuster, M.: Better generative models for sequential data problems: bidirectional recurrent mixture density networks. In: Leen, T.K., Dietterich, T.G., Tresp, V. (eds.) Advances in Neural Information Processing Systems 13 - Proceedings of the 2000 Conference, NIPS 2000, Denver, CO, United States, pp. 589–594. Neural Information Processing System Foundation (2000)
Sreehari: The Sign Language MNIST (2016). https://www.kaggle.com/datamunge/sign-language-mnist. Accessed 03 July 2020
Srivastava, N., Mansimov, E., Salakhutdinov, R.: Unsupervised learning of video representations using LSTMs. In: Bach, F., Blei, D. (eds.) 32nd International Conference on Machine Learning, ICML 2015, Lile, France, vol. 1, pp. 843–852. International Machine Learning Society (IMLS) (2015)
Sutskever, I., Vinyals, O., Le, Q.V.: Sequence to sequence learning with neural networks. In: Ghahramani, Z., Welling, M., Cortes, C., Lawrence, N., Weinberger, K. (eds.) Advances in Neural Information Processing Systems 27–28th Annual Conference on Neural Information Processing Systems 2014, NIPS 2014, Montreal, QC, Canada, vol. 4, pp. 3104–3112. Neural Information Processing Systems Foundation (2014)
van der Walt, S., Schönberger, J.: Scikit-Image: Image Processing in Python (2019). https://scikit-image.org/. Accessed 11 June 2020
Wang, Y., Tian, F.: Recurrent residual learning for sequence classification. In: Su, J., Duh, K., Carreras, X. (eds.) 2016 Conference on Empirical Methods in Natural Language Processing, EMNLP 2016, Austin, TX, United States, pp. 938–943. Association for Computational Linguistics (ACL) (2016)
Xiao, H., Rasul, K., Vollgraf, R.: Fashion-MNIST (2017). https://github.com/zalandoresearch/fashion-mnist. Accessed 01 June 2020
Acknowledgements
This work was supported by UNAM-PAPIIT IA103420. EJEC thanks the support of CONACyT.
Author information
Authors and Affiliations
Corresponding authors
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2020 Springer Nature Switzerland AG
About this paper
Cite this paper
Ek-Chacón, E., Molino-Minero-Re, E. (2020). LSTM Classification under Changes in Sequences Order. In: Martínez-Villaseñor, L., Herrera-Alcántara, O., Ponce, H., Castro-Espinoza, F.A. (eds) Advances in Soft Computing. MICAI 2020. Lecture Notes in Computer Science(), vol 12468. Springer, Cham. https://doi.org/10.1007/978-3-030-60884-2_1
Download citation
DOI: https://doi.org/10.1007/978-3-030-60884-2_1
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-030-60883-5
Online ISBN: 978-3-030-60884-2
eBook Packages: Computer ScienceComputer Science (R0)