A Tutorial on Fisher Information
By Alexander Ly, Maarten Marsman, Josine Verhagen, Raoul P. P. P. Grasman, Eric-Jan Wagenmakers. 2017
323 citations recorded by the source corpus.
In many statistical applications that concern mathematical psychologists, the concept of Fisher information plays an important role. In this tutorial we clarify the concept of Fisher information as it manifests itself across three different statistical paradigms. First, in the frequentist paradigm, Fisher information is used to construct hypothesis tests and confidence intervals using maximum likelihood estimators; second, in the Bayesian paradigm, Fisher information is used to define a default prior; lastly, in the minimum description length paradigm, Fisher information is used to measure model complexity.
Subfields: Artificial Intelligence, Statistical and Nonlinear Physics
View Paper: https://www.semanticscholar.org/paper/5fe1c0975f843167d20226c226a6eb6075359a82
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Artificial Intelligence
- Quoc V. Le
- Samuel L. Smith
- Richard Zhang
- Eli Shechtman
- Jascha Narain Sohl-Dickstein
- Yoshua Bengio
- Sergey Levine
- Trevor Darrell
In Statistical and Nonlinear Physics
- Richard Zhang
- Eli Shechtman
- Quoc V. Le
- Jascha Narain Sohl-Dickstein
- Ivor Wai-Hung Tsang
- Samuel L. Smith
- Jiangchao Yao
- Daniel Duckworth
Citations in the corpus
- Stochastic natural gradient descent draws posterior samples in function space (2018, methodology): “We pre-condition the gradient by the inverse Fisher information, Fx(ω) (defined in Section 4), which we estimate over the training inputs (Ly et al., 2017).”
- Degeneration in VAE: in the Light of Fisher Information Loss (2018, background): “(7), the term EZ [ ∂logF (x) ∂φl ] is proved zero in many works of information theory (Brunel and Nadal, 1998; Ly et al., 2017).”
- Few-shot Image Generation with Elastic Weight Consolidation (2020, methodology): “Recall that in mathematical statistics, the Fisher Information F could tell how well we estimate the model parameters given the observations [26].”