liveBook · Manning

Perturbation test

A perturbation test evaluates a system by applying a controlled change to its inputs, components, or conditions and observing the response.

How 2 Manning books explain it

Quantization and Fast Inference

Perturbation analysis evaluates how changing or disturbing one component affects a system's output or measured behavior.

How it is used
Ranks model layers by accuracy impact to identify where aggressive quantization risks disproportionate loss.
How deep it goes
Defines one-layer-at-a-time testing and automates comparisons against baseline accuracy, without detailing evaluation setup.
Examples
Quantizes selected layer weights with max-absolute scaling divided by 127 and measures the resulting accuracy drop.
What you can do after
Diagnose/debug with it by ranking layers according to accuracy loss after one-at-a-time weight quantization.

Read how Quantization and Fast Inference explains perturbation test →

Build AI Drug Discovery Pipelines

A perturbation experiment measures system responses after a controlled genetic, chemical, or physical intervention.

How it is used
Supports dynamic modeling of cellular responses and in-silico drug-discovery experiments.
How deep it goes
Defines intervention-response data and applications but stops short of experimental protocols or measurement procedures.
Examples
Uses E. coli growth-phenotype prediction, yeast models, mammalian pathway models, and virtual drug testing as examples.
What you can do after
Evaluate cellular and drug responses in computational models before some wet-lab experiments.

Read how Build AI Drug Discovery Pipelines explains perturbation test →

Also covered in

Related concepts

Questions about perturbation test

What is perturbation test?

A perturbation test evaluates a system by applying a controlled change to its inputs, components, or conditions and observing the response.

Which Manning books explain perturbation test?

Quantization and Fast Inference and Build AI Drug Discovery Pipelines.

What is the difference between perturbation test and data perturbation?

A perturbation test evaluates a system by applying a controlled change to its inputs, components, or conditions and observing the response. Data perturbation is the deliberate modification or randomization of data or released results to reduce disclosure of original values while retaining analytical utility.

These guides are generated from the text of Manning books by Manning's knowledge graph and link to the sections they draw on. The full books are available on liveBook and at manning.com.