Warning: Undefined variable $post in /home/quantdev/thechangelab.stanford.edu/wp-content/themes/lagunita-theme/Lagunita.php on line 124
Warning: Attempt to read property "ID" on null in /home/quantdev/thechangelab.stanford.edu/wp-content/themes/lagunita-theme/Lagunita.php on line 124
Variational Inference as Tractable Approximation for Bayesian Models of Psychological Change Processes
- |
- News & Updates
- |
- Conference
- |
- Variational Inference as Tractable Approximation for Bayesian Models of Psychological...
Large-scale Bayesian modeling of nonlinear psychological change processes can be intractable and unsustainable. Variational inference facilitates Bayesian analysis through estimation of approximate posterior distributions. For systems that can withstand the simplifying assumptions of variational inference, the time savings may be worth the small loss in posterior estimation accuracy. Future work will focus on determining which kinds of models variational inference is best suited to — those that maintain accuracy even after making the simplifying assumptions — and comparing the performance of variational algorithms of varying complexity.
![Honors Posters at Symposium of Undergraduate Research and Public Service [April 25, 2026]](https://thechangelab.stanford.edu/wp-content/uploads/2026/04/Garg_SURPSVig_20260425-300x95.png "Honors Posters at Symposium of Undergraduate Research and Public Service [April 25, 2026]")
