CS702(B) Deep & Reinforcement Learning complete study material for RGPV CSE 7th Semester. Download unit-wise notes, important questions, PYQ analysis and exam resources for Deep Learning, Autoencoders, CNN Architectures, Reinforcement Learning, Q-Learning, DQN, Policy Gradient, Actor-Critic and Advanced RL methods.
Deep & Reinforcement Learning combines neural network-based learning with decision-making methods. This subject covers deep neural networks, optimization methods, autoencoders, CNN architectures and reinforcement learning algorithms used for intelligent agents.
Study neurons, activation functions, gradient descent, RNN, LSTM, GRU and attention mechanisms.
Learn LeNet, AlexNet, VGGNet, GoogLeNet, ResNet and visualization methods.
Understand MDP, Bellman equation, value iteration, Q-learning, DQN and policy gradient.
Open any unit to access detailed notes, important questions and PYQ analysis.
History of deep learning, McCulloch Pitts neuron, threshold logic, activation functions, gradient descent, Adam, RNN, BPTT, GRU, LSTM, encoder-decoder and attention mechanism.
Autoencoders and PCA, denoising autoencoders, sparse autoencoders, contractive autoencoders, bias-variance tradeoff, L2 regularization, dropout and normalization.
Greedy layerwise pre-training, activation functions, weight initialization, word representations, CNN, LeNet, AlexNet, ZF-Net, VGGNet, GoogLeNet, ResNet and deep learning trends.
Introduction to RL, bandit algorithms, UCB, PAC, MDP, Bellman optimality, dynamic programming, value iteration, policy iteration, Q-learning and TD learning.
Fitted Q, Deep Q-Learning, DQN, policy gradient, actor-critic method, hierarchical RL, POMDPs, inverse reinforcement learning, maximum entropy IRL and GAIL.
Complete syllabus of CS702(B) Deep & Reinforcement Learning for RGPV CSE 7th Semester.
History of Deep Learning, McCulloch Pitts Neuron, Thresholding Logic, Activation Functions, Gradient Descent, Momentum Based GD, Nesterov Accelerated GD, Stochastic GD, AdaGrad, RMSProp, Adam, Eigenvalue Decomposition, RNN, BPTT, Vanishing and Exploding Gradients, Truncated BPTT, GRU, LSTMs, Encoder Decoder Models and Attention Mechanism.
Autoencoders and relation to PCA, regularization in autoencoders, denoising autoencoders, sparse autoencoders, contractive autoencoders, bias variance tradeoff, L2 regularization, early stopping, dataset augmentation, parameter sharing and tying, injecting noise, ensemble methods, dropout, batch normalization, instance normalization and group normalization.
Greedy layerwise pre-training, better activation functions, better weight initialization methods, learning vectorial representations of words, convolutional neural networks, LeNet, AlexNet, ZF-Net, VGGNet, GoogLeNet, ResNet, visualizing CNN, guided backpropagation, Deep Dream, Deep Art and recent trends in deep learning architectures.
Introduction to reinforcement learning, bandit algorithms, UCB, PAC, median elimination, policy gradient, full RL and MDPs, Bellman optimality, dynamic programming, value iteration, policy iteration, Q-learning, temporal difference methods, eligibility traces, function approximation and least squares methods.
Fitted Q, Deep Q-Learning, advanced Q-learning algorithms, learning policies by imitating optimal controllers, DQN and Policy Gradient, policy gradient algorithms for full RL, hierarchical RL, POMDPs, actor-critic method, inverse reinforcement learning, maximum entropy deep inverse reinforcement learning, generative adversarial imitation learning and recent trends in RL architectures.
Upload PDFs in the pdfs folder using these file names.
Most important questions for 7 marks and 14 marks answers.
| Topic | Unit | Expected Frequency | Importance |
|---|---|---|---|
| Activation Functions | Unit 1 | High | ⭐⭐⭐⭐ |
| Gradient Descent Optimizers | Unit 1 | Very High | ⭐⭐⭐⭐⭐ |
| RNN, BPTT, GRU, LSTM | Unit 1 | Very High | ⭐⭐⭐⭐⭐ |
| Autoencoders | Unit 2 | Very High | ⭐⭐⭐⭐⭐ |
| Regularization & Dropout | Unit 2 | High | ⭐⭐⭐⭐ |
| CNN Architectures | Unit 3 | Very High | ⭐⭐⭐⭐⭐ |
| Reinforcement Learning & MDP | Unit 4 | Very High | ⭐⭐⭐⭐⭐ |
| Bellman Optimality | Unit 4 | Very High | ⭐⭐⭐⭐⭐ |
| Q-Learning & TD Learning | Unit 4 | Very High | ⭐⭐⭐⭐⭐ |
| DQN, Policy Gradient, Actor-Critic | Unit 5 | Very High | ⭐⭐⭐⭐⭐ |
Deep Learning is a machine learning technique that uses multi-layer neural networks to learn complex patterns from data.
Reinforcement Learning is a learning method where an agent learns actions by interacting with an environment and receiving rewards.
Q-Learning is a reinforcement learning algorithm used to learn the best action-value function for decision-making.
DQN stands for Deep Q-Network. It combines Q-learning with deep neural networks.
Actor-Critic combines policy-based and value-based learning using an actor and a critic model.
Yes, it is scoring if you prepare definitions, algorithms, architecture diagrams and comparisons properly.
Architecture models, ADLs, CBAM, ATAM, ADD and documentation.
Open Software ArchitectureFuzzy systems, genetic algorithms, rough sets, HMM and swarm intelligence.
Open CIOLAP, classification, clustering, association rules, Apriori and FP Growth.
Open DMW