# @stem_antics on Instagram

- **Type:** Video
- **Original URL:** https://www.instagram.com/p/DbzmmQSgG_r
- **Gondola URL:** https://gondola.cc/posts/68785328-stem-antics-instagram
- **Thumbnail:** https://img.gondola.cc/tr:w-,h-,fo-auto/postThumbnails/3edda723bb.jpg
- **Posted:** 2026-08-09T04:46:30.000+00:00
- **Account Owner:** Stem Antics (@stem_antics) — https://gondola.cc/stem_antics

## Caption

A neural network that looks at an image one pixel at a time is doomed. A 224x224 colour photograph carries over 150,000 input values, and wiring every one to every neuron in a hidden layer produces billions of parameters before the model learns anything. Convolutional neural networks solve this by refusing to treat pixels as independent facts. They assume something about the world: that a pattern worth recognising in the top left corner is still worth recognising in the bottom right.

That assumption becomes an architecture.

- A kernel is a small grid of weights, typically 3x3 or 5x5, that slides across the image
- At each position it multiplies overlapping pixel values by its weights and sums them into a single number
- The result is a feature map, a new image showing where that pattern occurred
- The same kernel is reused everywhere, so one filter costs nine weights rather than 150,000

This is weight sharing, and it is the whole trick. Translation invariance is built into the structure rather than learned from data.

The kernels are not designed. They are initialised randomly and shaped by gradient descent until they become useful. What emerges is consistent and slightly eerie: first-layer filters converge on oriented edge detectors and colour blobs closely resembling the simple cells Hubel and Wiesel recorded in cat visual cortex in 1959. Deeper layers compose these primitives into corners, textures, then eyes, wheels, faces. Nobody specified that hierarchy. It falls out of stacking convolutions and letting the error signal propagate backwards.

Pooling layers then discard spatial precision deliberately, downsampling maps so the network cares that an edge exists rather than exactly where. Precision is traded for abstraction, layer by layer, until a few thousand numbers describe what a million pixels meant.

Intelligence often begins as a good assumption about the world, rigidly enforced.

#stemantics #machinelearning #computervision #deeplearning #neuralnetworks

## Stats

- **Views:** 4,602
- **Likes:** 157
- **Shares:** 0
- **Comments:** 3

## Tags

deeplearning, machinelearning, computervision, neuralnetworks, stemantics

---
Copyright (c) Gondola