liveBook · Manning

Auxiliary model

As explained in Machine Learning System Design

A secondary machine learning model is an additional learned model that provides an alternative or corrective output when a primary model is unsuitable.

An auxiliary model is a secondary learned model that supports a primary model by validating its data assumptions or supplying an alternative when its output is unsuitable. section 7.3.2 section 13.4

Use an auxiliary model to:

  • Compare labeled and unlabeled datasets: concatenate them, label the datasets, and train a binary classifier. ROC AUC near 0.5 suggests similarity; higher separability signals distribution differences. section 7.3.2
  • Provide a fallback when the primary model is unavailable, uncertain, outside an acceptable range, or affected by changing inputs. section 13.4
  • Correct a primary model with a dual-model setup: keep stable features in the first model and give the secondary model a larger feature set. section 13.4

Production fallback

An equipment-failure predictor can switch to a secondary model when the primary model fails, loses confidence, or encounters unacceptable input drift. Activation may be automatic, manual, or hybrid; a rule-based system, constant, or explicit error can also serve as the fallback. section 13.4

How Machine Learning System Design covers it

How deep it goes
Explains the binary-classification validation procedure and fallback choices, but stops short of implementation details.
Examples
Uses labeled-versus-unlabeled datasets, dual-model feature sets, and manufacturing equipment-failure prediction as concrete applications.
What it assumes
Assumes familiarity with machine learning models, datasets, classification, model outputs, and monitoring signals.
What you can do after
Evaluate dataset similarity or operate a primary model with a planned secondary prediction path.
How it connects
Connects auxiliary models to adversarial validation, binary classification, ROC AUC, concept drift, feature drift monitoring, baselines, and rule-based systems.
Cover of Machine Learning System Design
liveBookmanning.com
Read Machine Learning System Design on liveBook Get the book at manning.com

Related concepts

This explainer was generated by Manning's knowledge graph from the text of Machine Learning System Design. The section links point to the parts of the book it draws on.