Population dynamics: Variance and the sigmoid activation function

185 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Semi-analytical approximations to statistical moments of sigmoid and softmax mappings of normal variables (2017, background): “…(see, e.g., Wichmann and Hill, 2001), input/output relationships of artificial (see, e.g., Jain et al., 1996) and natural neurons (see, e.g., Marreiros et al., 2008), cumulative distribution functions of usual probability density functions (see, e.g., Waissi and Rossin, 1996), logistic…”
EEG and MEG Data Analysis in SPM8 (2011, methodology): “The second is SEP which uses a variant of this model; however, priors on the neuronal dynamics make them faster to model early evoked responses [60].”
Deep Learning on Image Denoising: An overview (2019, unknown): “The activation functions such as Sigmoid [139, 93] and Tanh [81, 47] can convert the linear input into non-linearity through W and b as follows.”
Dynamic causal modeling for EEG and MEG (2009, methodology): “Note that DCMs for M/EEG are necessarily complex, because they reﬂect the inherent complexity of the processes generating M/EEG data.”
Dynamic causal modelling: A critical review of the biophysical and statistical foundations (2011, methodology): “F with respect to q indirectly minimizes the Kullback-Leibler divergence between”
Dynamic models of large-scale brain activity (2017, background): “The breadth of this sigmoid function implicitly incorporates the variance of individual neural thresholds, as well as the dispersion of their states 47 .”