Write a paper (500 words) that explains how Neural Networks work. Your paper must address ALL of following questions. Your paper should be a cohesive piece of writing and should not consist of simply a list of answers to the questions.
Explain the process of training the computer to learn patterns found in images. Why is this process a data-driven approach? Discuss the terms: training set, the testing set, true labels or the ground truth.
What is a derivative? Explains and give a simple geometric interpretation. What is a gradient? Give a geometric interpretation of the gradient. Use the analogy from Lecture 4.
The main algorithm that makes Neural Networks capable of understanding data is the gradient descent algorithm. Explain this algorithm. How is the gradient used in the algorithm?