Selection through rejection: Reconsidering the invariant learning paradigm

Quarterly Journal of Experimental Psychology, 51A(1), 1-17, 1998
Selection through rejection: Reconsidering the invariant learning paradigm
Elizabeth Churchill, David Gilmore
Abstract

Two experiments are reported that investigate the nature of selections in the McGeorge and Burton (1990) invariant learning paradigm. McGeorge and Burton suggest that subjects implicitly acquire abstract knowledge of an invariant feature (usually the presence of the digit “3”) in a set of 30 stimuli.

McGeorge and Burton's analysis has recently been challenged by Cock, Berry, and Gaffan (1994) and by Wright and Burton (1995). In this paper, we demonstrate that performance is based on knowledge of other aspects of the learning set besides the invariant digit, but that this knowledge is still implicit.

Altering the nature of the learning stimuli to highlight these co-varying features enhances the effects and increases the reporting of explicit knowledge. Our results indicate that performance within this paradigm is more easily characterized as rejection of salient negatives than selection of positive instances, but that salience is not based simply on similarity.

Another publication from the same category: Machine Learning and Data Science

IEEE Computing Conference 2018, London, UK

Regularization of the Kernel Matrix via Covariance Matrix Shrinkage Estimation

The kernel trick concept, formulated as an inner product in a feature space, facilitates powerful extensions to many well-known algorithms. While the kernel matrix involves inner products in the feature space, the sample covariance matrix of the data requires outer products. Therefore, their spectral properties are tightly connected. This allows us to examine the kernel matrix through the sample covariance matrix in the feature space and vice versa. The use of kernels often involves a large number of features, compared to the number of observations. In this scenario, the sample covariance matrix is not well-conditioned nor is it necessarily invertible, mandating a solution to the problem of estimating high-dimensional covariance matrices under small sample size conditions. We tackle this problem through the use of a shrinkage estimator that offers a compromise between the sample covariance matrix and a well-conditioned matrix (also known as the "target") with the aim of minimizing the mean-squared error (MSE). We propose a distribution-free kernel matrix regularization approach that is tuned directly from the kernel matrix, avoiding the need to address the feature space explicitly. Numerical simulations demonstrate that the proposed regularization is effective in classification tasks.

Keywords