Abstract
We examine a Geometric Deep Learning model as a thermodynamic system treating the weights as non-quantum and non-relativistic particles. We employ the notion of temperature previously defined in [5] and study it in the various layers for GCN and GAT models. Potential future applications of our findings are discussed.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
References
Ackley, D.H., Hinton, G.E., Sejnowski, T.J.: A learning algorithm for Boltzmann machines. Cogn. Sci. 9(1), 147–169 (1985)
Barbaresco, F.: Lie group statistics and Lie group machine learning based on Souriau Lie groups thermodynamics and Koszul-Souriau-Fisher metric: new entropy definition as generalized Casimir invariant function in coadjoint representation. Entropy 22(6), 642 (2020)
Chaudhari, P., Soatto, S.: Stochastic gradient descent performs variational inference, converges to limit cycles for deep networks. In: 2018 Information Theory and Applications Workshop (ITA), pp. 1–10 (2018)
Chaudhari, P., et al.: Entropy-sgd: biasing gradient descent into wide valleys. In: International Conference on Learning Representations ICLR 2017, 1611.01838 (2017)
Fioresi, R., Faglioni, F., Morri, F., Squadrani, L.: On the thermodynamic interpretation of deep learning systems. In: Geometric Science of Information: 5th International Conference (2021)
Hopfield, J.J.: Neurons with graded response have collective computational properties like those of two-state neurons. PNAS 81(10), 3088–3092 (1984)
Jaynes, E.T.: Information theory and statistical mechanics. Phys. Rev. 106(4), 620 (1957)
Lapenna, M., Faglioni, F., Fioresi, R.: Thermodynamics modeling of deep learning systems. Front. Phys. 11 (2023). https://doi.org/10.3389/fphy.2023.1145156
Marle, C.-M.: From tools in symplectic and Poisson geometry to J. M. Souriau’s theories of statistical mechanics and thermodynamics. Entropy 18(10), 370 (2016)
Salakhutdinov, R., Hinton, G.: Deep Boltzmann machines. In: Artificial Intelligence and Statistics, pp. 440–455. PMLR (2009)
Anahory Simoes, A., De Leon, M., Lainz Valcazar, M., De Diego, D.M.: Contact geometry for simple the rmodynamical systems with friction. Proc. R. Soci. A 476(2241), 20200244 (2020)
Ioffe, S., Szegedy, C.: Batch normalization: accelerating deep network training by reducing internal covariate shift. In: Proceedings of the 32nd International Conference on Machine Learning, PMLR 37, pp. 448–456 (2015)
Monti, F., Boscaini, D., Masci, J., Rodolà, E., et al.: Geometric deep learning on graphs and manifolds using mixture model CNNs. In: 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (2017). https://doi.org/10.1109/CVPR.2018.00335
Gou, M., Xiong, F., Camps, O., Sznaier, M.: MoNet: moments embedding network. In: 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 1611.08402 (2018)
Avelar, P.H.C., Tavares, A.R., da Silveira, T.L.T., et al.: Superpixel image classification with graph attention networks. In: 33rd SIBGRAPI Conference on Graphics, Patterns and Images (SIBGRAPI), pp. 203–209 (2020). https://doi.org/10.1109/SIBGRAPI51738.2020.00035
Nair, V., Hinton, G.E.: Rectified linear units improve restricted Boltzmann machines. In: Proceedings of the 27th International Conference on Machine Learning (ICML-10), 21–24 June (2010)
Srivastava, N., Hinton, G., Krizhevsky, A., et al.: Dropout: a simple way to prevent neural networks from overfitting. J. Mach. Learn. Res. 15(1), 1929–1958 (2014)
Kipf, T.N., Welling, M.: Semi-supervised classification with graph convolutional networks. In: International Conference on Learning Representations ICLR 2017 (2017). https://openreview.net/forum?id=SJU4ayYgl
Risken, H.: The Fokker-Planck Equation. In: Methods of Solution and Applications, Springer, Berlin (1996). https://doi.org/10.1007/978-3-642-61544-3
Velikovi, B.Y., et al.: Graph attention networks. In: International Conference on Learning Representations ICLR 2018 (2018). https://doi.org/10.40550/ARXIV.1710.10903
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2023 The Author(s), under exclusive license to Springer Nature Switzerland AG
About this paper
Cite this paper
Lapenna, M., Faglioni, F., Zanchetta, F., Fioresi, R. (2023). Geometric Deep Learning: A Temperature Based Analysis of Graph Neural Networks. In: Nielsen, F., Barbaresco, F. (eds) Geometric Science of Information. GSI 2023. Lecture Notes in Computer Science, vol 14072. Springer, Cham. https://doi.org/10.1007/978-3-031-38299-4_65
Download citation
DOI: https://doi.org/10.1007/978-3-031-38299-4_65
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-031-38298-7
Online ISBN: 978-3-031-38299-4
eBook Packages: Computer ScienceComputer Science (R0)