Variational Inference - Interactive Visualization

Variational inference approximates intractable posteriors with a simpler distribution q(z;φ). We optimize φ to maximize the ELBO = E_q[log p(x,z)] - KL(q||p). The ELBO is tractable and acts as a lower bound on log p(x). This visualization shows q(z) (Gaussian) moving to approximate a complex target p(z|x), with ELBO increasing as KL decreases.

• Watch Gaussian q(z) optimize toward a complex target distribution
• See ELBO increasing over optimization iterations
• Understand ELBO = reconstruction term - KL(q||p)
• Compare diagonal Gaussian vs full covariance approximation quality
• Foundation for VAEs, AEVB algorithm, and scalable Bayesian inference

Part of the EngineersOfAI Interactive 3D - free interactive visualizations covering every major concept in machine learning and AI engineering. Hover any element for a plain-English explanation. No code required.