Skip to main content
Glite Dictionary
🧠 A conclusion 🤖 Model prediction
inference

inference

8 5.7
Using a computer program that learned from examples to make predictions for new data
  • noun
  • /ˈɪnfərəns/
  • Specialized
translation icon : inferencia
  • We are currently exploring joint inference models and identification of relations that span multiple sentences using coreference resolution.
  • Bayesian inference
  • Statistical inference
  • low inference

A neural network does not do inference on the fly.

Autoplay Next

Examples

  • A neural network does not do inference on the fly.

    Lex Fridman

  • Each closed-form inference model is also trained on the full training set.

    Academic text (2017)

  • On the engineering side, anywhere where you have noise, inference comes in.

  • We develop multiple models of inference to address this step.

    Academic text (2017)

  • The first and most important point is that WatsonPaths has the ability to engage in inference.

    Academic text (2017)

  • His research focuses on deep learning in natural language processing and probabilistic joint inference.

    Academic text (2017)

  • David W. Buchanan is interested in joint inference problems in large AI systems, especially using Bayesian approaches.

    Academic text (2017)

  • To perform inference, we use Metropolis-Hastings sampling over a factor graph representation of the inference graph.

    Academic text (2017)

  • Once the inference graph has been built, we run a probabilistic inference engine over the graph to generate new confidences.

    Academic text (2017)

  • Using a trained algorithm, the model generates predictions through inference on new data.

Surface Forms

inference singular
inferences plural

Morphology

inference = infer (semi-transparent) = infer + ence

Morphologically regular derivation from 'infer', but the computing/ML sense ('running inference') is a technical extension that may not be obvious to general learners.

Etymology

Inference comes from the Latin parts in- 'into' and ferre 'to carry'. In computing, an inference is when a trained model 'carries' what it has learned to new data to make a prediction, so 'running inference' is simply using the model to give answers for new cases.