.compute_uncertainty

Given a fitted Keras submodel and covariate input <code>x</code>, compute uncertainty
estimates according to the <code>uncertainty_method</code>.<ul>
<li><code>"epistemic"</code>: estimates only epistemic variance (via MC Dropout passes).</li>
<li><code>"aleatoric"</code>: uses deterministic quantile heads to estimate aleatoric variance.</li>
<li><code>"both"</code>: combines aleatoric and epistemic using variance decomposition.</li>
<li>Otherwise: returns <code>NA</code> placeholders.</li>
</ul>

internal

Neural Additive Model framework based on Generalized Additive Models from Hastie & Tibshirani (1990, ISBN:9780412343902), which trains a different neural network to estimate the contribution of each feature to the response variable. The networks are trained independently leveraging the local scoring and backfitting algorithms to ensure that the Generalized Additive Model converges and it is additive. The resultant Neural Network is a highly accurate and interpretable deep learning model, which can be used for high-risk AI practices where decision-making should be based on accountable and interpretable algorithms.

Ines Ortega-Fernandez

neuralGAM

Interpretable Neural Network Based on Generalized Additive
Models

Marta Sestelo

.compute_uncertainty function

<dl><dt>model</dt>
<dd>Fitted Keras model for a single smooth term.</dd>
<dt>x</dt>
<dd>Input covariate matrix (or vector; will be reshaped as needed).</dd>
<dt>uncertainty_method</dt>
<dd>Character; one of <code>"epistemic"</code>, <code>"aleatoric"</code>, <code>"both"</code>, or <code>"none"</code>.</dd>
<dt>alpha</dt>
<dd>Coverage level (e.g. 0.05 for 95% bands).</dd>
<dt>forward_passes</dt>
<dd>Integer; number of MC Dropout passes.</dd></dl>

Arguments

Ines Ortega-Fernandez, Marta Sestelo

Author

Internal helper: compute uncertainty decomposition (epistemic / aleatoric / both) — .compute_uncertainty

<dl>

<dt>model</dt>
<dd>Fitted Keras model for a single smooth term.</dd>


<dt>x</dt>
<dd>Input covariate matrix (or vector; will be reshaped as needed).</dd>


<dt>uncertainty_method</dt>
<dd>Character; one of <code>"epistemic"</code>, <code>"aleatoric"</code>, <code>"both"</code>, or <code>"none"</code>.</dd>


<dt>alpha</dt>
<dd>Coverage level (e.g. 0.05 for 95% bands).</dd>


<dt>forward_passes</dt>
<dd>Integer; number of MC Dropout passes.</dd>

</dl>

.compute_uncertainty: Internal helper: compute uncertainty decomposition (epistemic / aleatoric / both)

Description

Usage

Value

Arguments

Author