Tutorial on maximum likelihood estimation

By In Jae Myung. 2003

1,815 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Control and Systems Engineering

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

In Control and Systems Engineering

Citations in the corpus

A survey of data fusion in smart city applications (2019, methodology): “Common techniques under this category are Maximum Like-lihood [62], Kalman Filter [63], Particle Filter [64], and Covariance Consistency Model [65].”
Probing Representation Forgetting in Supervised and Unsupervised Continual Learning (2022, methodology): “The importance of the parameters is approximated via the diagonal of the Fisher information matrix [32].”
Stochastic mechano-chemical kinetics of molecular motors: A multidisciplinary enterprise from a physicist’s perspective (2012, methodology): “ML approach [452] is based on finding the estimates of the set of model parameters that corresponds to the maximum of the likelihood P (~ d|~ m) for a fixed set of data ~ d.”
On a signal detection approach to m-alternative forced choice with bias, with maximum likelihood and Bayesian approaches to estimation (2012, methodology): “(4) is that, as for the yes/no situation, it is a simple probit model, and so it can easily be fit with standard software (e.g., DeCarlo, 2003) using maximum likelihood estimation (see Myung, 2003).”
Bayesian Cognitive Modeling: A Practical Course (2014, background): “Note that with a uniform prior, the mode of a posterior distribution coincides with the classical maximum likelihood estimate or MLE , θ̂ = k/n = 0.9 (Myung, 2003).”
Self-exciting point process models of civilian deaths in Iraq (2011, methodology): “For simple models with a small number of parameters, it is often possible to calculate the maximum likelihood explicitly ( Myung, 2003 ).”
IRGAN: A Minimax Game for Unifying Generative and Discriminative Information Retrieval Models (2017, methodology): “ng problems that estimate the data p.d.f. p„x”and supervised learning problems that estimate the conditional p.d.f. p„yjx”, maximum likelihood estimation (MLE) plays as the standard learning solution [30]. In IR, MLE is also widely used as an estimation method for many relevance features or retrieva”
A Tutorial on Fisher Information (2017, methodology): “Consequently, for vector-valued parameters θ, this information25 lower bound does not inform us, whether we should stop our search for a better estimator.26 Both the Hájek-LeCam convolution theorem and the Cramér-Fréchet-Rao information lower27 bound are statements conditioned on the data being gene”
Using Group-Based Trajectory and Growth Mixture Modeling to Identify Classes of Change Trajectories (2016, background): “Local maxima are an artifact of maximum likelihood estimation procedures and suggest that the best log likelihood was not reached and therefore the best fitting parameters were not found (Myung, 2003).”
Is there a credibility crisis in strategic management research? Evidence on the reproducibility of study findings (2017, methodology): “There are no known solutions for this problem, but one way to address it is to use a range of randomly generated initial parameters (Myung, 2003), which is the procedure implemented by IBM SPSS Amos, LISREL, and many other programs typically used by strategic management researchers.”
Cognitive model decomposition of the BART: Assessment and application (2011, methodology): “The likelihood of the parameters given the data is proportional to the probability of the data given the parameters (e.g., Edwards, 1992; Myung, 2003), so that L(α, μ, γ , β|D) ∝ p(D|α, μ, γ , β).”
Audiovisual synchrony and temporal order judgments: Effects of experimental method and stimulus type (2008, methodology): “To test whether the correlation between SJ PSS values is a consequence of determining the PSS by averaging the synchrony boundaries, SJ2 and SJ3 PSS values were also determined from the mean of individually fitted truncated Gaussians (see Fujisaki et al., 2004) using maximum-likelihood estimation ( ”
LEMNA: Explaining Deep Learning based Security Applications (2018, methodology): “Different from a standard linear regression, our optimization objective is intractable andwe cannot simply utilizeMLE to perform minimization.”
How to use MATLAB to fit the ex-Gaussian and other probability functions to a distribution of response times (2008, methodology): “A detailed mathematical tutorial on likelihood methods is presented in Myung (2003).”
Nonlinear Regression Models and Applications in Agricultural Research (2015, methodology): “…method can affect the parameter estimates (Ruppert et al., 1989), but in general, estimates from least squares and maximum likelihood methods tend to differ only when the data are not normally distributed and are approximately identical when the data follow a normal distribution (Myung, 2003).”