There are various feature selection processes used in Machine Learning like Filter methods, wrapper methods or embedded methods.
In case of wrapper method the feature selection is dependent upon the algorithm. Wrapper method uses greedy strategy, assessing all potential feature combinations against the evaluation criterion
Wrapper methods, on average, have a higher predictive accuracy than filter approaches.
Forward feature selection, backward feature elimination, exhaustive feature selection and recursive feature elimination are a few wrapper method techniques.
Filter methods select the relevant methods and do not interact with the classifier to name a few filter methods are ANOVA, Chi-Square, Fisher’s score.
Embedded methods are a combination of filter and wrapper methods. Lasso Regularization, Elastic Net are a few embedded methods.
It depends upon the methods used whether feature selection will be dependent or independent on the algorithm.