Beginner3Blue1Brown
What a neural network is
Grant Sanderson walks through a network that reads handwritten digits. No code. The point is the structure: layers of numbers, weights, and a final guess.
But what is a neural network? | Deep learning chapter 1
Key points
The running example is handwritten digits. The input is a 28×28 grid of pixel values, 784 numbers.
Neurons sit in layers. Each neuron holds an activation: a number computed from the previous layer.
A connection has a weight, and a neuron has a bias. The weighted inputs are summed, then passed through a function. The video contrasts ReLU and sigmoid.
The whole network is one function. In this example it has on the order of 13,000 weights and biases, and 10 output numbers, one per digit.
Learning, in this video, means changing those weights and biases. You do not write a separate rule for each way someone draws a 3.
This video does not train the network, derive backpropagation, or discuss language models, price, or products.
Sign in to earn points for it.
Ask while you watch
The coach stays on this lesson’s notes. If the notes do not cover it, the answer says so.
Sign in to ask and add it to the thread. Sign in
Questions on this lesson
Anyone can read. Posting needs a session.
No questions yet. Be the first to ask.