Most Common Interview questions for a Deep Learning Engineer - PART 5/5

These are some of the most common interview questions for Deep Learning Engineer
PART 5/5

1. What is RNN?
2. When RNN and when CNN?
3. What are GRU?
4. What is LSTM?
5. What are forget and update gates in LSTM?
6. What is BiRNN?
7. What are word embeddings?
8. What is Beam Search?
9. What is Bleu Score?
10. What are Attention Models and how do we build them?
11. What are Auto-Encoders?
12. What is Binary Cross Entropy?
13. What are the different Classification Losses?
14. What is Negative Log Likelihood?
15. What is Margin Loss?
16. What is Soft Margin Loss?
17. What are the different Regression Losses?
18. What is L1 Loss?
19. What is MSE loss?
20. What is KL Divergence?
21. What is GAN's?
22. What are Adversarial Networks?
23. How do we do weights reuse in GAN's?
24. What is the Discriminator and Generator Loss in a GAN?
25. Why is a CNN better than Dense Layers?
26. How is the computation efficiency of CNN compared to a Simple NN?
27. How do we calculate the number of learnable parameters in the network?
28. What is Tensorboard? Why is it useful?
29. What is Grid Search?
30. Is it necessary to have the activation functions differentiable?

The Best Books recommended to improve your Machine Learning skills:

   

The links to the initial 4 parts are here:

PART 1

PART 2

PART 3

PART 4

Hope you have liked the Series on Deep Learning Interview Questions. Let me know your views in the comments below.

For Machine Learning Interview Questions check this series: Machine Learning

#MachineLearning #DeepLearning #ArtificialIntelligence #AI

Most Common Interview questions for a Deep Learning Engineer - PART 4/5

These are some of the most common interview questions for Deep Learning Engineer
PART 4/5

1. Given 16x16x3 input with a 25 filters of 3x3x3 what is the output shape with stride = 1, padding = 'valid'?
2. What is a Pooling Layer? What are the types?
3. Does Pooling Operation reduce the spatial dimension or the depth?
4. What is Average Pooling?
5. What is Max Pooling?
6. Why Convolution Operation?
7. Explain any Basic CNN based architectures?
8. Does VGG network have 3x3 or 5x5 kernel?
9. What is ResNet architecture?
10. What are DenseNets?
11. What is Inception Network?
12. What are Skip Connections?
13. What are 1x1 Convolutions? Where do we use them and what are the benefits of it?
14. What is Emsembling?
15. What is Multicrop at Test time method do?
16. What is Object Localisation?
17. What is Object Detection?
18. What is a Sliding Window technique?
19. What is Landmark detection?
20. What is YOLO algorithm?
21. What is an SSD algorithm?
22. What is Intersection Over Union?
23. What is Non-Max Suppression?
24. What are Anchor Boxes in YOLO?
25. What is RCNN algorithm?
26. What is RCNN, Fast RCNN, Faster RCNN?
27. What is Siamese Network?
28. What is Triplet Loss Function in Face Recognition?
29. What is Neural Style Transfer?
30. Can Convolutions (CNN) operations be used in 1D or 2D cases?

The Best Books recommended to improve your Machine Learning skills:

   

For Part 5 of the series click here: PART 5

Most Common Interview questions for a Deep Learning Engineer - PART 3/5

These are some of the most common interview questions for Deep Learning Engineer
PART 3/5

1. How do we go with selecting these hyperparameters?
2. What is Batch Normalization?
3. What are the parameters involved in Batch Normalization?
4. Are there any learnable parameters in Batch Normalization step?
5. Why Batch Normalization?
6. What is covariance shift?
7. How do we do Batch Normalization at Train and Test time?
8. What are the benefits of Batch Normalization?
9. What is Multiclass Classification?
10. What is a Sigmoid Layer?
11. What is a Softmax function?
12. When do we do softmax or sigmoid activation?
13. What is Cross Entropy?
14. What is L1 regularization?
15. What is L2 regularization?
16. Which is better L1 or L2 regularization?
17. What is precision/Recall/F1 score? Explain with an example?
18. How do we do Error Analysis in ML?
19. What is Ceiling Analysis?
20. What is Transfer Learning?
21. What is Fine Tuning?
22. What is Multitask learning? How do we do that?
23. What is End-to-End Deep Learning?
24. What is CNN?
25. What is Translation Invariance?
26. What is Convolution operation, explain with example?
27. What is Edge detection?
28. What is a Sobel operator?
29. What is padding?
30. What is Valid/Same padding? Which is better?

The Best Books recommended to improve your Machine Learning skills:

   

For Part 4 of the series click here: PART 4

Most Common Interview questions for a Deep Learning Engineer - PART 2/5

These are some of the most common interview questions for Deep Learning Engineer
PART 2/5

1. How do we overcome underfitting and overfitting?
2. Does more data solve high variance or bias problem?
3. What is Regularization?
4. What are the types of regularisation?
5. What is the learning rate in NN?
6. is it better to have a high or low learning rate?
7. Does increasing the learning rate get you results faster?
8. What are dropouts?
9. What do we do with dropouts in Training/Testing phase?
10. Can you give an intuition of dropouts?
11. What is data augmentation?
12. What are the different types of Data Augmentation?
13. What is Early stopping? How do we achieve that?
14. What framework is better Tensorflow/Pytorch/any other or better to code it yourself as it have better control over the results?
15. What is gradient checking? How do we do that?
16. Is it required to Normalize the training set ? if so why and explain the process.?
17. Is it required to Normalise the Test set? If so why and explain the process.
18. What is vanishing and exploding gradients?
19. What are the Optimization Algorithms in ML?
20. What is Batch Gradient Descent?
21. What is Mini-Batch Gradient Descent?
22. What is Stochastic Gradient Descent?
23. What is Gradient Descent with Momentum? What are the hyperparameters involved in it? Explain?
24. What is RMSProp? What are the hyperparameters involved in it?
25. What is Adam Optimization?
26. What is Adagrad?
27. What is Learning Rate Decay?
28. What is a saddle point in Loss landscape?
29. How does Adam Optimization help?
30. What are the different hyperparameters involved in a NN?

The Best Books recommended to improve your Machine Learning skills:

   

For Part 3 of the series click here: PART 3

Most Common Interview questions for a Deep Learning Engineer - PART 1/5

These are some of the most common interview questions for Deep Learning Engineer
PART 1/5

1. What is a Linear Regression? Answer
2. What is Logistic Regression?
3. What is the Cost function used for a Regression problem?
4. What is the cost function used for a Classification Problem?
5. What is a Neural Network?
6. What is Computational Graph?
7. What is Forward Pass and Backward Pass in a NN?
8. What is Gradient Descent?
9. What is a rank 1 in a numpy array?
10. What are the activation functions?
11. Name and explain different activation functions.
12. Why do we need non-linearity in a network?
13. Why is tanh better than sigmoid?
14. Why is relu better than others?
15. What is relu activation?
16. What is leaky relu? 
17. How do you write a relu/leaky relu function?
18. How do we initialize the weights in a network?
19. what is Xavier initialization?
20. Why "Deep" Neural Networks?
21. How do we proceed with an ML workflow?
22. How do we split the data in ML workflow?
23. What is cross-validation?
24. Is it necessary to have Dev and Test set form the same distribution?
25. What is Bias and Variance in ML?
26. Is good to have High Bias or High Variance :)
27. How do we overcome High Bias?
28. How do we overcome High Variance?
29. What is the Optimal Bayes Error?
30. Is underfitting High/Low - Bias or Variance Situation?

The Best Books recommended to improve your Machine Learning skills:

   

For Part 2 of the series click here: PART 2

C++ HelloWorld code

This is a simple C++ HelloWorld code:

For running it on Visual Code check out this video:
https://www.youtube.com/watch?v=DIw02CaEusY&vl=en

#include <iostream>

using namespace std;

int main(){

cout << "Hello World" << endl;

return 0;

}