Artificial intelligence

Abstract: The article considers the method of factor cluster analysis which allows automatically retrain the onboard recognition system of an unmanned aerial system. The task of informational synthesis of an on-board system for identifying frames is solved within the information-extreme intellectual technology of data analysis, based on maxi- mizing the informational ability of the system during machine learning. Based on the functional approach to modeling cognitive processes inherent to humans during forming and making classification decisions, it was proposed a categorical model in the form of a direct graph. According to this model, the algorithmic support of the information extreme factor cluster analysis is developed. It allows automatically retrain the system when expanding the alphabet of recognition classes. According to this algorithm, the on-board recognition system preliminarily carries out the information-extremal machine learning of recognition classes of relatively low power. When new classes appear, their unclassified structured recognition attribute vectors form additional learning matrixes. After reaching a representational volume, additional learning matrix joins the input learning matrix and the on-board recognition system is retrained. Forming additional learning matrixes of new recognition classes is carried out by the agglomerative algorithm of cluster analysis of unclassified vectors by k-means clustering. As a criterion of optimizing machine-learning parameters, we used the modified Kullback criterion which is a functional of the exact characteristics of classification solutions. To increase the functional efficiency of factor cluster analysis, it is proposed to increase the depth of machine learning by optimizing the parameters of image processing frames.

Keywords: Information-extreme machine learning; Identification; Digital image frame of the region

References:

1. Naumenko I., Myronenko M., Piatachenko V. (2019). Information-extreme learning of on-board system for recognition of ground vehicle. CEUR Workshop Proceedings, 121-132.
2. Dovbysh, A.S., Martynenko, S.S., Kovalenko, A.S., Budnyk, M.M. (2011). Information-extreme algorithm for recognizing current distribution maps in magnetocardiography. Journal of Automation and Information Sciences, 63-70.
3. Dovbysh, А.S., Moskalenko, V.V., Rizhova, A.S. (2016). Information-Extreme Method for Classification of Observations with Categorical Attributes. Cibernetica and Systems Analysis, 52(2), 45-52.
4. Schmidt J. (2019). Recent advances and applications of machine learning in solid-state materials science. NPG Computational Materials, 5(1), 1–36.
5. Nguyen T.D., Kwoh C.K. (2016). Efficient agglomerative hierarchical clustering for biological sequence analysis. IEEE Region 10 Annual International Conference, Proceedings/TENCON. Institute of Electrical and Electronics Engineers Inc., Vol. 2016-January.
6. Snoek J., Larochelle H., Adams R.P. Practical Bayesian Optimization of Machine Learning Algorithms.
7. Ahalya G., Pandey H.M. (2015). Data clustering approaches survey and analysis. 2015 International Conference on Futuristic Trends on Computational Analysis and Knowledge Management, Noida 532-537.
8. Protsenko O., Savchenko T., Myronenko M., Prikhodchenko O. (2020). Informational and extreme machine learning for onboard recognition system of ground objects. IEEE 11th International Conference on Dependable Systems, Services and Technologies, 213-218.