The effect of noise on a class of energy-based learning rules
A Bazzani1, D Remondini, N Intrator
1Department of Physics, University of Bologna, 40126, Bologna, Italy. bazzani@bologna.infn.it
Abstract:
Westudy the selectivity properties of neurons based on BCM and kurtosis energy functions in a general case of noisy high-dimensional input space. The proposed approach, which is used for characterization of the stable states, can be generalized to a whole class of energy functions. We characterize the critical noise levels beyond which the selectivity is destroyed. We also perform a quantitative analysis of such transitions, which shows interesting dependency on data set size. We observe that the robustness to noise of the BCM neuron (Bienenstock, Cooper, & Munro, 1982; Intrator & Cooper, 1992) increases as a function of dimensionality. We explicitly compute the separability limit of BCM and kurtosis learning rules in the case of a bimodal input distribution. Numerical simulations show a stronger robustness of the BCM rule for practical data set size when compared with kurtosis.
Related Concept Videos
Sound Intensity Level
The human ear can perceive an extensive range of sound intensity, necessitating the use of the logarithmic scale to define a physical quantity—the intensity level. It is a ratio of two intensities and hence a...
Sound Waves: Interference
Classification of Signals
A continuous-time signal holds a value at every instant in time, representing information seamlessly. In contrast, a discrete-time signal holds values only at specific moments, often denoted as x(n), where...
Energy and Power Signals
Introduction to Learning
In contrast to learned behaviors, unlearned behaviors such as crying, sexual...
Real-World Application of Classical Conditioning
Higher-order, or second-order, conditioning occurs when a neutral stimulus becomes associated with an already established conditioned stimulus through repeated pairings. For instance, if a dog has been...


