<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>VBF Experiments, May 2026 on mlbot.blog</title><link>https://mlbot.blog/series/vbf-experiments-may-2026/</link><description>Recent content in VBF Experiments, May 2026 on mlbot.blog</description><generator>Hugo</generator><language>en-US</language><lastBuildDate>Sat, 02 May 2026 04:56:46 +0530</lastBuildDate><atom:link href="https://mlbot.blog/series/vbf-experiments-may-2026/index.xml" rel="self" type="application/rss+xml"/><item><title>Strict Nonlinear Filtering With Mixtures, Particles, And Flows</title><link>https://mlbot.blog/posts/strict-nonlinear-filtering-mixtures-particles-flows/</link><pubDate>Sat, 02 May 2026 04:30:00 +0530</pubDate><guid>https://mlbot.blog/posts/strict-nonlinear-filtering-mixtures-particles-flows/</guid><description>&lt;p&gt;A Kalman filter is a beautiful bargain. If the dynamics and observations are
linear and Gaussian, the filtering distribution stays Gaussian forever:&lt;/p&gt;
\[
\begin{aligned}
q^F_t(z_t) &amp;= p(z_t \mid y_{1:t}) \\
&amp;= \mathcal{N}(m_t, P_t).
\end{aligned}
\]&lt;p&gt;Each update carries only a mean and variance. The filter is online, cheap, and
exact.&lt;/p&gt;</description></item></channel></rss>