keyword
small sample size effects
Small sample size effects refer to the statistical biases, performance degradation, and estimation errors that occur in pattern recognition and machine learning when the volume of available data is too small relative to the number of features or the complexity of the model. When training data is limited, parameter estimates become unstable and classifiers are prone to overfitting, often leading to phenomena where adding descriptive features paradoxically worsens classification accuracy. In addition, an insufficient number of test samples makes it difficult to reliably assess true model performance, frequently resulting in highly variable or optimistically biased error estimates during feature selection and system validation.
1 item
