Deep Learning Interview Prep Course | Full Course [100 Q&A’s]

Q1 – What is Deep Learning?
Q2 – What is Deep Learning?
Q3 – What is a Neural Network?
Q4 – Explain the concept of a neuron in Deep Learning.
Q5 – Explain architecture of Neural Networks in simple way
Q6 – What is an activation function in a Neural Network?
Q7 – Name few popular activation functions and describe them
Q8 – What happens if you do not use any activation functions in a NN?
Q9 – Describe how training of basic Neural Networks works
Q10 – What is Gradient Descent?
Q11 – What is the function of an optimizer in Deep Learning?
Q12 – What is backpropagation, and why is it important in Deep Learning?
Q13 – How is backpropagation different from gradient descent?
Q14 – Describe what Vanishing Gradient Problem is and it’s impact on NN
Q15 – Describe what Exploding Gradients Problem is and it’s impact on NN
Q16 – There is a neuron results in a large error in backpropagation. Reason?
Q17 – What do you understand by a computational graph?
Q18 – What is Loss Function and what are various Loss functions used in DL?
Q19 – What is Cross Entropy loss function and how is it called in industry?
Q20 – Why is Cross-entropy preferred as cost function for multi-class classification?
Q21 – What is SGD and why it’s used in training Neural Networks?
Q22 – Why does stochastic gradient descent oscillate towards local minima?
Q23: How is GD different from SGD
Q24: What is SGD with Momentum
Q25 – Batch Gradient Descent, Minibatch Gradient Descent vs SGD
Q26: What is impact of Batch Size
Q27: Batch Size vs Model Performance
Q28: What is Hessian, usage in DL
Q29: What is RMSProp and how does it work?
Q30: What is Adaptive Learning
Q31: What is Adam Optimizer
Q32: What is AdamW Algorithm in Neural Networks
Q33: What is Batch Normalization
Q34: What is Layer Normalization
Q35: What are Residual Connections
Q36: What is Gradient Clipping
Q37: What is Xavier Initialization
Q38: What are ways to solve Vanishing Gradients
Q39: How to solve Exploding Gradient Problem
Q40: What is Overfitting
Q41: What is Dropout
Q42: How does Dropout prevent Overfitting in Neural Networks
Q43: Is Dropout like Random Forest
Q44: What is the impact of DropOut on the training vs testing
Q45: What are L2 and L1 Regularizations for Overfitting NN
Q46: What is the difference between L1 and L2 Regularisations
Q47: How do L1 vs L2 Regularization impact the Weights in a NN?
Q48: What is the Curse of Dimensionality in Machine Learning | Deep Learning Interview Question
Q49 – How Deep Learning models tackle the Curse of Dimensionality | Deep Learning Interview Question
Q50: What are Generative Models, give examples?
Q51 – What are Discriminative Models, give examples?
Q52 – What is the difference between generative and discriminative models?
Q53 – What are Autoencoders and How Do They Work?
Q54: What is the Difference Beetween Autoenconders and other Neural Networks?
Q55 – What are some popular autoencoders, mention few?
Q56 – What is the role of the Loss function in Autoencoders, & how is it different from other NN?
Q57 – How do autoencoders differ from (PCA)?
Q58 – Which one is better for reconstruction linear autoencoder or PCA?
Q59 – How can you recreate PCA with neural networks?
Q60 – Can You Explain How Autoencoders Can be Used for Anomaly Detection?
Q61 – What are some applications of AutoEncoders
Q62 – How can uncertainty be introduced into Autoencoders, & what are the benefits and challenges of doing so?
Q63 – Can you explain what VAE is and describe its training process?
Q64 – Explain what Kullback-Leibler (KL) divergence is & why does it matter in VAEs?
Q65 – Can you explain what reconstruction loss is & it’s function in VAEs?
Q66 – What is ELBO & What is this trade-off between reconstruction quality & regularization?
Q67 – Can you explain the training & optimization process of VAEs?
Q68 – How would you balance reconstruction quality and latent space regularization in a practical Variational Autoencoder implementation?
Q69 – What is Reparametrization trick and why is it important?
Q70 – What is DGG “Deep Clustering via a Gaussian-mixture Variational Autoencoder (VAE)” with Graph Embedding
Q71 – How does a neural network with one layer and one input and output compare to a logistic regression?
Q72 – In a logistic regression model, will all the gradient descent algorithms lead to the same model if run for a long time?
Q73 – What is a Convolutional Neural Network?
Q74 – What is padding and why it’s used in Convolutional Neural Networks (CNNs)?
Q75 – Padded Convolutions: What are Valid and Same Paddings?
Q76 – What is stride in CNN and why is it used?
Q77 – What is the impact of Stride size on CNNs?
Q78 – What is Pooling, what is the intuition behind it and why is it used in CNNs?
Q79 – What are common types of pooling in CNN?
Q80 – Why min pooling is not used?
Q81 – What is translation invariance and why is it important?
Q82 – How does a 1D Convolutional Neural Network (CNN) work?
Q83 – What are Recurrent Neural Networks, and walk me through the architecture of RNNs.
Q84 – What are the main disadvantages of RNNs, especially in Machine Translation Tasks?
Q85 – What are some applications of RNN?
Q86 – What technique is commonly used in RNNs to combat the Vanishing Gradient Problem?
Q87 – What are LSTMs and their key components?
Q88 – What limitations of RNN that LSTMs do and don’t address and how?
Q89 – What is a gated recurrent unit (GRU) and how is it different from LSTMs?
Q90 – Describe how Generative Adversarial Networks (GANs) work and the roles of the generator and discriminator in learning.
Q91 – Describe how would you use GANs for image translation or creating photorealistic images?
Q92 – How would you address mode collapse and vanishing gradients in GAN training, and what is their impact on data quality?
Q93- Minimax and Nash Equilibrium in GAN
Q94 – What are token embeddings and what is their function?
Q95 – What is self-attention mechanism?
Q96 – What is Multi-Head Self-Attention and how does it enable more effective processing of sequences in Transformers?
Q97 – What are transformers and why are they important in combating problems of models like RNN and LSTMs?
Q98 – Walk me through the architecture of transformers.
Q99 – What are positional encodings and how are they calculated?
Q100 – Why do we add positional encodings to Transformers but not to