At its core, machine learning is about . We build a model, make predictions, calculate how wrong those predictions are (the "loss"), and then adjust the model to make it better.
dJdwthe fraction with numerator d cap J and denominator d w end-fraction tells us how the cost changes if we tweak the weight 2. Partial Derivatives and Gradients calculus for machine learning pdf link
Without calculus, we would be guessing blindly. With calculus, we have a roadmap to the best solution. At its core, machine learning is about