Which statements are correct regarding optimizers in machine learning $1($a$)$ Batch gradient descent algorithm computes exact gradients before updating model parameters in each step. $($b$)$ Adam optimizer uses both the momentum method and adaptive learning rate concept. $($c$)$ AdaGrad does not use an adaptive learning rate $($d$)$ Stochastic gradient descent computes gradients approximately during each iteration, and hence, there may be oscillations near the minimum.