An Adaptive Multipath Linear Interpolation Method for Sample Optimization

When using machine learning methods to make predictions, the problem of small sample sizes or highly noisy observation samples is common. Current mainstream expansion cannot handle data noise well. We propose a multipath method (AMLI) based on idea linear interpolation, which mainly solves insufficient prediction size large error between observed and actual distribution. The rationale AMLI divide original feature space into several subspaces with equal samples, randomly extract from each subspace as class, then perform interpolation in same class (i.e., K-path interpolation). After processing, valid are greatly expanded, structure adjusted, average reduced so that effect model improved. hyperparameters this have an intuitive explanation usually require little calibration. compared proposed variety demonstrated can significantly improve result. also plus classes by combining clustering present theoretical proofs effectiveness methods.