Meridian Signal LabMeridian Signal LabDSP / VISION / ML
DOC · MERIDIAN SIGNAL LAB · EST. 2015

Hard AI ideas, drawn out until they make sense.

An independent research-and-education studio. We take one idea at a time — the Fourier transform, motion detection, overfitting — and diagram it the way it should have been explained the first time.

FIG. 00 — SIGNAL PATHSchematic of a waveform flowing through nodes into a neural graph and a lensSCALE 1:1
§1.0

Fields of study

NODE / DSP

Signal Processing

Filtering, sampling, spectral analysis — and why the Fourier transform keeps turning up everywhere.

NODE / CV

Computer Vision

From convolution to modern video analytics: tracking, motion, and the trade-offs no abstract mentions.

NODE / ML

Applied ML

What a model is really doing, where it breaks, and how to tell a genuine result from a lucky one.

§2.0

Register — latest entries

SIG-01
2026-05-12
Signal Processing

The Fourier transform, explained the way it finally clicked

From “a signal is a sum of waves” to why the frequency domain makes half of engineering easier.

The Fourier transform, explained the way it finally clicked
CV-02
2026-03-28
Computer Vision

How video analytics actually detects motion

Background subtraction, optical flow, and the gotchas of detection on real, messy footage.

How video analytics actually detects motion
ML-03
2026-02-04
Applied ML

Overfitting is not a bug you fix once

Why the gap between training and reality is the whole job, plus a checklist to catch it early.

Overfitting is not a bug you fix once
Full register →
Most technical writing is a 40-page paper for six experts, or a blog post that skips the part you got stuck on. We work the honest middle — careful enough to be correct, patient enough to be useful.
// MERIDIAN SIGNAL LAB — EDITORIAL CHARTER

We keep one local key to remember your cookie choice. Cookie notice.