Theme: calculation of spectral coefficients of signals using the machine learning method based on xaar bases

AS A RESULT OF IMPROVING THE HAAR BASES BASED ON MACHINE LEARNING, THE RESULTS OF CALCULATING THE SPECTRAL COEFFICIENTS.

	Function	piece - invariant				piece - linear				piece-quadratic
	Number of signals	64		128		64		128		64		128
№		Ks	Nk %	Ks	Nk %	Ks	Nk %	Ks	Nk %	s	Nk %	Ks	Nk %
1.	Y= e x	1	38	1	45	1.7	65.3	4.1	5.7	16	96	43	99.2
2.	Y=e 2 x COS (4x)	1	28	1	33	1.6	47.2	2.8	65.1	11	7 0	9	87.3
3.	th (x)	1	35	1	42	1.1	67	2.3	73	13	92	16	94.5

From experience received in the table results [0; 1] into 64, 128 pieces separate received values to the function when we put harvest has been values .

CALCULATION OF SPECTRAL COEFFICIENTS OF GEOPHYSICAL SIGNALS USING IMPROVED HAAR BASES.

	Function	piece - invariant				piece - linear				piece-quadratic
	Number of signals	64		128		64		128		64		128
№
		Ks	Nk %	Ks	Nk %	Ks	Nk %	Ks	Nk %	Ks	Nk %	Ks	Nk %
1	X B	1.7	46.5	2.6	61	2.7	63.4	3.2	68.5	3.7	73.4	4.7	78.5
2	MX B	1.8	47.6	2.6	62.3	2.9	65.5	3.36	0.1	5.8	82.8	6.4	85.1

The above table shows the results of calculating the spectral coefficients of geophysical signals based on the XB - Haar bases. MXB – Results from Improving Haar Databases Based on Machine Learning.

CONCLUSION

CONCLUSION

Signals spectral coefficients in the calculation Haar piece-invariant , piece- linear and piece-quadratic bases analysis done. Haar bases convenience coefficients count for fast algorithms existence. But many practical issues in solving this bases method accuracy enough not Haar piece-polynomial bases teaching the way with accuracy increase possible analysis done. Analysis in the process Haar piece-quadratic bases choose received and 64, 128 values for calculations implemented . Based on Haar bases in Table 1 received results . In Table 2, Haar bases from improvement harvest done results . To tables attention giving if we In Table 1, y = e x function out of 128 values for when used zero coefficients are 88.2%. In Table 2, the function y=e x is achieving 99.2% when using 128 values. In table 4, the spectral coefficients of geophysical signals were calculated and increased by 78.5% - 85.1%. It can be seen that the accuracy of the values obtained as a result of model improvement is high.