Calculus is a branch of mathematics that focuses on the concepts of derivatives and integrals, which are techniques that describe how functions change.
Calculus, linear algebra and probability are the foundations of Machine Learning. Learning about these topics will give us a deeper understanding of how models might work by definition and even allow us to develop new ones.
The derivatives (derivative) measure the rate of change. If we think of a function that describes the position of a moving object with respect to time, the derivative of that function would give us the velocity of the object at any given time.
The integrals (integrals) give us an accumulation of quantities. Continuing with the example of the moving object, if we had a function that describes the velocity of the object with respect to time, the integral of that function would tell us how far the object has traveled in a time interval.
Machine learning uses derivatives in optimization problems. Optimization algorithms such as gradient descent use derivatives to decide whether to increase or decrease weights to maximize or minimize some objective (e.g., the accuracy of a model or error functions). Derivatives also help us approximate nonlinear functions to linear functions with constant slopes. Thus, by having a constant slope, the model can calibrate its weights (increase or decrease them) to approach the target value.