Answer / Neelabh Shukla

1. Stochastic Gradient Descent (SGD) is a variation of Gradient Descent where instead of computing gradients over an entire dataset, it computes gradients for a single example from the dataset at each iteration. This makes SGD computationally efficient but prone to noisy updates due to its high variance. 2. Batch Gradient Descent (BGD) is another variation where gradients are computed over the entire training set in one go. This helps reduce the noise compared to SGD, but it can be slower and require more memory. 3. Mini-Batch Gradient Descent (MBGD) is a compromise between SGD and BGD. In MBGD, gradients are computed for a small batch of examples (mini-batch) instead of the entire dataset or a single example. This reduces the noise while still maintaining computational efficiency.

What will happen if the learning rate is set too low or too high?

1 Answers  

---

What are the differences between Machine Learning and Deep Learning?

1 Answers  

---

What is Bagging and Boosting?

1 Answers  

---

Explain gradient descent?

1 Answers  

---

Can relu function be used in output layer?

1 Answers  

---

What is the use of leaky relu function?

1 Answers  

---

Please explain what is deep learning?

1 Answers  

---

What do you mean by deep learning and why has it become popular now?

1 Answers  

---

What is an auto-encoder?

1 Answers  

---

What is data normalization?

1 Answers  

---

Why is zero initialization not a good weight initialization process?

1 Answers  

---

What do you understand by boltzmann machine?

1 Answers  

---