What general term describes the failure mode in which a model's deployment input distribution differs from its training distribution, degrading performance, and which sequential imitation learners induce on themselves?

Show answer

Distribution shift

Distribution shift (the input-only case is called covariate shift) occurs whenever deployment data is drawn from a different distribution than training data. In sequential decision problems it is self-induced: a policy's own actions steer it into states its training data never covered.

Also asked as:

What term refers to the discrepancy between training and test data distributions that sequential decision-making agents often induce on themselves through their own actions?

Would you remember this in 6 months?

Trivia Forge uses spaced repetition so facts stick. Free, no signup needed.

Turn on Machine Learning practice →

More Machine Learning trivia

What heuristic is commonly used in K-means clustering to determine the optimal number of clusters by looking for a 'kink' in the cost function?
What is the common name for the regularization technique that adds the squared magnitude of coefficients as a penalty term to the loss function?
What is the computational complexity of the self-attention mechanism in a standard Transformer relative to the sequence length L?
Which SVM kernel is defined by K(x, x′) = exp(−γ‖x − x′‖²)?
What is the name for the empirical power-law relationships showing that a language model's loss falls predictably as model size, dataset size, and training compute increase?
What is the name of the alignment method that replaces many human preference labels with model self-critique and revision guided by a written set of principles?

See all 98 →