🔍 A distinguisher 🤖 A classifier

discriminator

8.1

A computer model that learns to tell real from fake data

noun
/dɪsˈkrɪmɪneɪtər/

Specialized

: discriminador

In a Generative Adversarial Network, the role of the discriminator is to identify whether an image is real or fake.

Examples

If it manages to produce an image that the discriminator can't tell is fake, then that's really good.
So trying to fool the discriminator, right?
And they do that essentially using a a clustering algorithm where the discriminator learns to assign the objects to groups.
The discriminator helps improve the quality of generated data by distinguishing it from authentic data.
To train the model effectively, the discriminator must learn to recognize subtle differences in the input data.

Synonyms

differentiator

vsdiscriminator

3 7.2

A quality that makes one product, person, or idea different

learns to separate authentic from synthetic data specifically

recognizer

vsdiscriminator

Specialized

1 6.3

A person or device that identifies things from patterns such as faces and handwriting

focuses on telling real data from generated data through classification

discerner

vsdiscriminator

1 7.5

A person who can notice small differences and good quality in things

operates automatically by training on examples to judge authenticity

Surface Forms

discriminator singular

discriminators plural

Morphology

discriminator = discriminate (semi-transparent) = discriminate + or

Morphologically it's the agentive '-or' on 'discriminate', but the specific technical interpretation in ML ('a model that distinguishes real from fake') requires domain knowledge beyond the base verb.

Etymology

Discriminator in machine learning fits its roots: dis- 'apart' and crimin- 'separate', so it is the part that 'separates' real data from fake data. A helpful way to remember it is to think of the model as a judge that learns to decide which examples are real and which are generated.