| rln-param | R Documentation |
These parameters are used with regularization learning networks (RLN), such
as tabular_rln() when fit with the brulee engine.
penalty_average(range = c(-15, -5), trans = scales::transform_log10())
step_rate(range = c(0, 8), trans = scales::transform_log10())
penalty_type(values = values_penalty_type)
values_penalty_type
range |
A two-element vector holding the defaults for the smallest and largest possible values, respectively. If a transformation is specified, these values should be in the transformed units. |
trans |
A |
values |
A character vector of possible values. |
penalty_average(): The target geometric mean of the per-weight
regularization coefficients (Theta in Shavitt and Segal (2018)). Best tuned
on the log10 scale.
step_rate(): The step size used to update the per-weight regularization
coefficients (nu in Shavitt and Segal (2018)). Best tuned on the log10
scale.
penalty_type(): The type of regularization norm applied to per-weight
coefficients. L1 is recommended by the original paper.
For penalty_average() and step_rate(), the value is passed to
brulee::brulee_rln() on the natural scale.
Shavitt, I., & Segal, E. (2018). Regularization learning networks: Deep learning for tabular datasets. Advances in Neural Information Processing Systems, 31, 1379-1389.
penalty_average()
step_rate()
values_penalty_type
penalty_type()
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.