Enhancing neural network applications in phonetic classification through population-based incremental learning
DOI:
https://doi.org/10.38152/bjtv7n4-027Abstract
Significant progress has been achieved in the field of Automatic Speech Recognition (ASR) thanks to the development of powerful algorithms. Research based on simplified biological models has led to the emergence of a new class called Estimation of Distribution Algorithms (EDA), which preserves significant partial solutions. Population-Based Incremental Learning (PBIL) is a technique that combines stochastic search and optimization. It is a statistical approach with evolutionary computation similar to EDAs, aimed at adapting recognition systems based on artificial Neural Networks (NN). Our main contribution lies in the improvement achieved by PBIL, which outperforms Genetic Algorithms (GA) by 16% and Evolutionary Strategies (ES) by 9.93%.
References
KA de Jong. Learning with genetic algorithm: an overview. Machine learning, vol.3, pp.121-138, 1988.
Larraňaga, P; Lozano, JA. A Review of estimation distributions algorithms. Chapter 3, Doct, Thesis, Univ of the Basque Country, 2002.
Očenášek, J. Parallel of estimation of distribution algorithm. Doctoral thesis, Brno university of Technology, 2001.
Baluja, S. Population based incremental leaning. Technical Report CMU-CS-94-163, Computer Science Department, Carnegie Mellon University, Pittsburg, PA, USA, 1994.
Shapiro, JL. The sensitivity of PBIL to the learning rate, and how detailed balance can remove It.Dept. Computer Science. Univ. Manchester, 2002.
Cuthbert, J; Braun, RM. Multi-H modulation indexes selection using the PBIL algorithm. Univ Cape Town, Conference Publication, No.0-7803-3019-6, IEEE, 1996.
Seiffert, U. MLP Training using Gas. ESANN'2001, Adelaide knowledge-based intelligent eng, System Center (KES), Australia, 2001.
Belew, R; Mc Inerney, J; Scharaudolph, NN. Evolving networks using the GAs with connectionist learning. CSE Tech, Rep.CS90-174, Computer Science, UCSD, 1990.
H. Zhu et al., « Masksleepnet: A cross-modality adaptation neural network for heterogeneous signals processing in sleep staging », IEEE J. Biomed. Health Inform., vol. 27, no 5, p. 2353‑2364, 2023.
M. Ghorvei, M. Kavianpour, M. T. Beheshti, et A. Ramezani, « Spatial graph convolutional neural network via structured subdomain adaptation and domain adversarial learning for bearing fault diagnosis », Neurocomputing, vol. 517, p. 44‑61, 2023.
Linde, Y; Buzo, A; Gray, RM. An algorithm for vector quantized design, IEEE Tran, Comp., 28(1), 1980.
Tlemsani, R; Neggaz, N; Benyettou, A; Dehak, N. A neuro-evolutionary model applied to phonetic classification. CNIE'04, Conf, Nat Electronic Engineering, pp.55, Univ. USTO Oran –Algeria, 22-23 Nov, 2004.
Benyettou, A; Neggaz, N; Tlemsani, R. Evolutionary algorithms based neural networks training for phonetic classification. ACIT04, Int, Conf Information Technology, pp. 468-472, Mentouri Univ. Constantine, Algeria, December 12-15, 2004.
Tlemsani, R; Neggaz, N; Benyettou. Improvement of the learning of RNs by evolutionary algorithms. 3rd Int. Conf, Sciences of Electronic, Technology of Information and Telecomm, SETIT05, Tunisia, March 27-31, 2005.
R. Zhou, S. Zhao, M. Luo, X. Meng, J. Ma, et J. Liu, « MFCC based real-time speech reproduction and recognition using distributed acoustic sensing technology », Optoelectron. Lett., vol. 20, no 4, p. 222‑227, avr. 2024, doi: 10.1007/s11801-024-3167-5.
M. Mohan, P. Dhanalakshmi, et R. S. Kumar, « Speech emotion classification using ensemble models with MFCC », Procedia Comput. Sci., vol. 218, p. 1857‑1868, 2023.
