Mean squared error
Mean squared error is a specific type of loss function. Mean square error is calculated by the average, specifically the mean, of errors that have been squared from data as it relates to a function ( often a regression line).
The utility of mean square error comes from the fact that squared numbers are positive, and that errors are squared before they are averaged. Although the absolute value of errors is also always positive, the mean square error has unique characteristics.
By using the square of errors, all values for error are translated into positive numbers. Data with a lot of variances that have values higher and lower than those determined by a function may have a mean error which underestimates the variance or even shows a mean error of zero, as positive and negative values can sum to zero. In order to get a better picture of the variance and bias in data sets, the mean squared error is often used as a measure.
Mean squared error and the root of mean squared error, can for many data sets show variance better than a simple absolute value of errors.
Where n is the number of values in a data set and we presume the error to be the distance from a function representing true values, the mean squared error can be calculated by 1/n (Σ errors^{2} ).
Mean squared errors are useful not only in machine learning, but many problems where there is a need to quantify the amount of lack of precision through statistical tools.
Related Radiopaedia articles
Artificial intelligence
 artificial intelligence (AI)
 imaging data sets
 computeraided diagnosis (CAD)
 natural language processing
 machine learning (overview)
 visualizing and understanding neural networks
 common data preparation/preprocessing steps
 DICOM to bitmap conversion
 dimensionality reduction
 scaling
 centering
 normalization
 principal component analysis
 training, testing and validation datasets
 augmentation
 loss function

optimization algorithms
 ADAM
 momentum (Nesterov)
 stochastic gradient descent
 minibatch gradient descent

regularisation
 linear and quadratic
 batch normalization
 ensembling
 rulebased expert systems
 glossary
 activation function
 anomaly detection
 automation bias
 backpropagation
 batch size
 computer vision
 concept drift
 cost function
 confusion matrix
 convolution
 cross validation
 curse of dimensionality
 dice similarity coefficient
 dimensionality reduction
 epoch
 explainable artificial intelligence/XAI
 feature extraction
 federated learning
 gradient descent
 ground truth
 hyperparameters
 image registration
 imputation
 iteration
 jaccard index
 linear algebra
 noise reduction
 normalization
 R (Programming language)
 radiomics quality score (RQS)
 Python (Programming language)
 segmentation
 semisupervised learning
 synthetic and augmented data
 overfitting
 underfitting
 transfer learning