Artificial neuron
An artificial neuron is the fundamental computing unit in a neural network. It is loosely inspired by biological neurons in the brain, though the resemblan...
How an Artificial Neuron Works
An artificial neuron is the fundamental computing unit in a neural network. It is loosely inspired by biological neurons in the brain, though the resemblance is mostly mathematical rather than biological. Each artificial neuron receives one or more numerical inputs, multiplies each input by a corresponding weight, sums the results, adds a bias term, and then passes the total through an activation function to produce an output.
Formally, if a neuron has inputs x1, x2, ..., xn with weights w1, w2, ..., wn and bias b, its output is f(w1*x1 + w2*x2 + ... + wn*xn + b), where f is the activation function. The weights determine how much influence each input has, while the bias shifts the activation threshold. Both weights and bias are learned during training.
A concrete example helps clarify this. Suppose a neuron is deciding whether a pixel is bright. It receives a pixel intensity value between 0 and 1 as input, multiplied by a learned weight. If the weighted sum exceeds a threshold set by the bias, the neuron fires a high output. Over training, the weight and bias adjust until the neuron reliably responds to the feature it is meant to detect.
The power of a neural network comes from combining thousands or millions of such neurons, each learning a slightly different aspect of the data. Individually, a single neuron can only solve linearly separable problems. Together in layers, they approximate arbitrarily complex functions.
Exam tip: when asked to describe an artificial neuron, always write out the mathematical formula with weights, inputs, bias, and activation function. Many candidates forget the bias term. Showing the formula and explaining that both weights and bias are trainable parameters demonstrates a thorough understanding that sets you apart.