this is from machine learning course. please help me solve this

Exercise 17points) 11. Explain the idea of gradient descent algorithm, define the cost function and the rule of updating coeficient of new repression 11. Explain why the cast function et logistic repression can be estimated by the tuliowing formation E(0)=-15 Slogh,(") + (1 - log(i-*") 1. Assuming at a certain iteration of sath gradient descent, we reach the global maximum value of the cost function EO is perfectly. What will happen nestand why? The algorithm stars at the global maximum -The algorithm updates the coefficient toward the minimum value of cost function is the answer remaining the same it we are using mini-batch gradient descent or stochastic-gradient descent