Abstract
Quaternion, as a hypercomplex number with three imaginary elements, is effective in characterizing three- and four-dimensional vector signals. Quaternion neural networks with randomly generated quaternions as the hidden node parameters become attractive for the good learning capability and generalization performance. In this paper, a novel incremental quaternion random neural network trained by extreme learning machine (IQ-ELM) is proposed. To fully exploit the second-order Q-properness statistic of quaternion random variables, the augmented quaternion vector is further applied in IQ-ELM (IAQ-ELM) for hypercomplex data learning. The network is constructed by gradually adding the hidden neuron one-by-one, where the output weight is optimized by minimizing the residual error based on the fundamental of the generalized HR calculus (GHR) of quaternion variable function. Experiments on multidimensional chaotic system regression, aircraft trajectory tracking, face and image recognition are conducted to show the effectiveness of IQ-ELM and IAQ-ELM. Comparisons to two popular quaternion RNNs, Schmidt NN (SNN) and random vector functional-link net (RVFL), are also provided to show the feasibility and superiority of using quaternions in RNN for incremental learning.
This work was supported by the National Natural Science Foundation of China (U1909209), the National Key Research and Development Program of China (2021YFE0100100, 2021YFE0205400), the Natural Science Key Foundation of Zhejiang Province (LZ22F030002), and the Research Funding of Education of Zhejiang Province (GK228810299201).
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
References
Javidi, S., Took, C.C., Mandic, D.P.: Fast independent component analysis algorithm for quaternion valued signals. IEEE Trans. Neural Netw. 22(12), 1967–1978 (2011)
Took, C.C., Mandic, D.P.: The quaternion LMS algorithm for adaptive filtering of hypercomplex processes. IEEE Trans. Signal Process. 57(4), 1316–1327 (2009)
Ujang, B.C., Took, C.C., Mandic, D.P.: Quaternion-valued nonlinear adaptive filtering. IEEE Trans. Neural Netw. 22(8), 1193–1206 (2011)
Jahanchahi, C., Mandic, D.P.: A class of quaternion Kalman filters. IEEE Trans. Signal Proce. 58(7), 3895–3901 (2010)
Tobar, F.A., Mandic, D.P.: Quaternion reproducing kernel Hilbert spaces: existence and uniqueness conditions. IEEE Trans. Infor. Theory 60(9), 5736–5749 (2014)
Arena, P., Fortuna, L., Re, R., et al.: Multilayer perceptions to approximate quaternion valued function. Int. J. Neural Syst. 6(4), 435–446 (1995)
Greenblatt, A.B., Agaian, S.S.: Introducing quaternion multi-valued neural networks with numerical examples. Inf. Sci. 423, 326–342 (2017)
Gaudet, C., Maida, A.: Deep quaternion networks. CoRR , arxiv.org/abs/1712.04604 (2018)
Zhu, X., Xu, Y., Xu, H., Chen, C.: Quaternion convolutional neural networks. CoRR, arxiv.org/abs/1903.00658 (2019)
Schmidt, W., Kraaijveld, M., Duin, R.: Feedforward neural networks with random weights. In: Proceedings of 11th IAPR International Conference on Pattern Recognition, vol. II. pp. 1–4 (1992)
Pao, Y., Park, G., Sobajic, D.: Learning and generalization characteristics of random vector functional-link net. Neurocomputing 6, 163–180 (1994)
Zhang, Y., Wu, J., Cai, Z., Philip, B., Yu, S.: An unsupervised parameter learning model for RVFL neural network. Neural Netw. 112, 85–97 (2019)
Huang, G., Chen, L., Siew, C.K.: Universal approximation using incremental constructive feedforward networks with random hidden nodes. IEEE Trans. Neural Netw. 17(4), 879–892 (2006)
Cao, J., Zhang, K., Yong, H., Lai, X., Chen, B., Lin, Z.: Extreme learning machine with affine transformation inputs in an activation function. IEEE Trans. Neural Netw. Learning Syst. 30(7), 2093–2107 (2019)
Yang, L., Song, S., Li, S., Chen, Y., Huang, G.: Graph embedding-based dimension reduction with extreme learning machine. IEEE Trans. Syst. Man. Cybern. Syst. (2019). https://doi.org/10.1109/TSMC.2019.2931003.
Chen, H., Wang, T., Cao, J., Vidal, P.-P., Yang, Y.: Dynamic quaternion extreme learning machine. IEEE Trans. Circ. Syst. II Express Briefs 68(8), 3012–3016 (2021)
Cao, J., Dai, H., Lei, B., Yin, C., Zeng, H., Kummert, A.: Maximum correntropy criterion-based hierarchical one-class classification. IEEE Trans. Neural Netw. Learn. Syst. 32(8), 3748–3754 (2021)
Lai, X., Cao, J., Lin, Z.: An accelerated maximally split ADMM for a class of generalized ridge regression. IEEE Trans. Neural Netw. Learn. Syst. (2021). https://doi.org/10.1109/TNNLS.2021.3104840
Minemoto, T., Isokawa, T., Nishimura, H., Siew, C.K.: Feed forward neural network with random quaternionic neurons. Signal Process. 136, 59–68 (2017)
Zhang, H., Wang, Y., Xu, D., Wang, J., Xu, L.: The augmented complex-valued extreme learning machine. Neurocomputing 311(15), 363–372 (2018)
Xu, D., Xia, Y., Mandic, D.P.: Optimization in quaternion dynamic systems: gradient, Hessian, and learning algorithms. IEEE Trans. Neural Netw. Learning Syst. 27(2), 249–261 (2016)
Xu, D., Mandic, D.P.: The theory of quaternion matrix derivatives. IEEE Trans. Signal Process. 63(6), 1543–1556 (2015)
Mandic, D.P., Jahanchahi, C., Took, C.C.: A quaternion gradient operator and its applications. IEEE Signal Process. Lett. 18(1), 47–50 (2011)
Took, C.C., Mandic, D.P.: Augmented second-order statistics of quaternion random signals. Signal Process. 91(2), 214–224 (2011)
Via, J., Ramirez, D., Santamaria, I.: Properness and widely linear processing of quaternion random vectors. IEEE Trans. Inf. Theory 56(7), 3502–3515 (2010)
Xiang, M., Kanna, S., Mandic, D.P.: Performance analysis of quaternion- valued adaptive filters in nonstationary environments. IEEE Trans. Signal Process. 66(6), 1566–1579 (2018)
Zhang, H., Lv, H.: Augmented quaternion extreme learning machine. IEEE Access 7, 90842–90850 (2019)
Huang, G., Li, M., Chen, L., Siew, C.K.: Incremental extreme learning machine with fully complex hidden nodes. Neurocomputing 71(4–6), 576–583 (2008)
Sudbery, A.: Quaternionic analysis. Math. Proc. Camb. Philos. Soc. 85(2), 199–225 (1979)
Ell, T.A., Sangwine, S.J.: Quaternion involutions and anti-involutions. Comput. Math. with Appl. 53(1), 137–143 (2007)
Xu, D., Jahanchahi, C., Took, C.C., Mandic, D.P.: Enabling quaternion derivatives: the generalized HR calculus. Roy. Soc. Open Sci. 2(8), 1–24 (2015)
Rumelhart, D.E., and McClelland, J.L.: Parallel Distributed Processing: Explorations in the Microstructure of Cognition, vol. 1, p. 567. MIT Press (1986)
Rössler, O.E.: An equation for continuous chaos. Phys. Lett. A 57(5), 397–398 (1976)
Matsumoto, T.: A chaotic attractor from Chua’s circuit. IEEE Trans. Circuits Syst. 31(12), 1055–1058 (1984)
Chen, G., Ueta, T.: Yet another chaotic attractor. Int. J. Bifurc. Chaos. 9(7), 1465–1466 (1999)
Chen, A., Lu, J., Lü, J., Yu, S.: Generating hyperchaotic Lü attractor via state feedback control. Stat. Mech. Appl. 364, 103–110 (2006)
Spacek, L.: Description of the collection of facial images (2011). http://cswww.essex.ac.uk/mv/allfaces/index.html
Acknowledgment
This work was supported by the National Natural Science Foundation of China (U1909209), the National Key Research and Development Program of China (2021YFE0100100, 2021YFE0205400) and the Open Research Projects of Zhejiang Lab (2021MC0AB04).
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
A Appendices
A Appendices
1.1 A.1 Proof of Proposition 1
With the well established GHR of quaternion variable function and Lemmas 2\({\sim } 4\), the optimization of (7) can be solved by finding the gradient of the objective to \(\beta _n\). It is observed that the norm of the objective residual error \(||e_n||^2\) can be equivalently calculated as \(e_n e_n^*\) or \(e_n^* e_n \), where \(e_n^*\) is the conjugate of \(e_n\). With the left- and right-derivatives, and the product rule in quaternion derivative, one can readily find that the same results will be obtained either on \(e_n e_n^*\) or \(e_n^* e_n \). We take the derivative to \(e_n^* e_n \) as an example in this paper, with Lemma 3, which can be expressed as
Since \(e_{n-1}\) and \(\beta _n\) are not related, \(\frac{\partial e_{n-1}}{\beta _n}=0\). With (4), we have
Here, \(e_{n-1}\) denotes the residual error of previous \(n-1\) nodes, \(\sigma _n\) represents the n-th hidden node output, \(S(\sigma _n)\) and \(V(\sigma _n)\) are the real (scalar) and imaginary parts of the output \(\sigma _n\), respectively. When \(\frac{\partial ||e_n||^2}{\partial \beta _n}=0\), the objective (7) reaches the minimum, so we have
If and only if \(\beta _n=\frac{e_{n-1}{\sigma ^*_n}}{||\sigma _n||^2}\), the objective (7) reaches the minimum. That finishes the proof.
Rights and permissions
Copyright information
© 2023 The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.
About this paper
Cite this paper
Cui, X., Wang, T., Chen, H., Lei, B., Vidal, PP., Cao, J. (2023). Incremental Quaternion Random Neural Networks. In: Sun, F., Cangelosi, A., Zhang, J., Yu, Y., Liu, H., Fang, B. (eds) Cognitive Systems and Information Processing. ICCSIP 2022. Communications in Computer and Information Science, vol 1787. Springer, Singapore. https://doi.org/10.1007/978-981-99-0617-8_13
Download citation
DOI: https://doi.org/10.1007/978-981-99-0617-8_13
Published:
Publisher Name: Springer, Singapore
Print ISBN: 978-981-99-0616-1
Online ISBN: 978-981-99-0617-8
eBook Packages: Computer ScienceComputer Science (R0)