Where the Output Layer Delivers the Answer
A free lesson from Swipe AI · Deep Learning & Neural Networks, level 2 · updated October 2026
By the Swipe team, founded by Colin · how we make lessons
The output layer is the network's final layer — the one whose values are actually read as the answer. Depending on the task, that might be a classification (which category something belongs to) or a prediction (a specific numeric estimate). Unlike hidden layers, the output layer's size is not really a free design choice either — it is fixed by however many answers the task requires.
A network sorting images into, say, five possible categories typically ends with five output units, one per category, each indicating how strongly the network favors that category. A network predicting a single number, like an estimated price, typically ends with just one output unit holding that number. The task decides the shape of the answer, and the output layer is built to match.
Structurally, the output layer works the same way any other layer does: it takes values from the previous layer and combines them using its own weights. What makes it different is only its position — it is the last step — and the fact that its size is dictated by the problem rather than chosen freely, the way a hidden layer's size might be.
Task: sort a photo into one of 3 animal categories Output layer: 3 units Unit 1 (cat) = 0.7 Unit 2 (dog) = 0.2 Unit 3 (bird) = 0.1 Highest value wins: the network's answer is cat
What determines how many units are in a network's output layer?
The output layer produces the network's final result, such as a ____ or a numeric prediction.
The output layer works fundamentally differently from every other layer, using entirely different math.
Input, Hidden, and Output Layers
10 questions · pass with 60% — just like in the app.
Which of the following best describes what the input layer does?
The number of units in the input layer is fixed by the shape of the data, not freely chosen.
A layer that is neither the input layer nor the output layer is called a ____ layer.
Why are hidden layers called 'hidden'?
Hidden layers are where most of a network's transformation work actually happens.
The ____ layer is the network's last layer, producing the final classification or prediction.
What fixes the number of units in a network's output layer?
The output layer uses fundamentally different math than the other layers in the network.
A network sorts photos into one of four categories. How many output units would it typically have?
Input, hidden, and output are all names for a layer's ____ in the network, not different kinds of math.
Free ai tools
Keep your streak going in Swipe AI
Swipe through every level of Deep Learning & Neural Networks on your iPhone — with XP, stars, streaks, a personal advisor and the full track beyond these free levels.
Get the Swipe apps →