Home › AKTU PYQ › Deep Learning › Important Topics

Deep Learning (BAI701) important topics for AKTU – unit-wise

Topics are ranked by how many times AKTU asked them across 5 papers. Finish the "very likely" topics first, then the top topics of each unit.

High-probability Deep Learning topics

  • Training a Network (Loss, Backprop, SGD) (Unit 1: Introduction to Neural Networks) – asked 10 times in 2023, 2025, 2026
  • Optimization in Deep Learning (Unit 4: Optimization & Generalization) – asked 10 times in 2023, 2025, 2026
  • Convolutional Networks (Unit 2: Deep Networks) – asked 9 times in 2023, 2025, 2026

Most asked Deep Learning topics in each unit

  • Unit 1 (Introduction to Neural Networks): Training a Network (Loss, Backprop, SGD) – 10 times; Linear Models (SVM, Perceptron, Logistic Regression) – 8 times; Neural Networks & Shallow Networks – 5 times
  • Unit 2 (Deep Networks): Convolutional Networks – 9 times; History & Theory of Deep Learning – 8 times; Regularization & Batch Normalization – 4 times
  • Unit 3 (Dimensionality Reduction): Linear Dimensionality Reduction (PCA, LDA) – 7 times; ConvNet Architectures (AlexNet, VGG, ResNet) – 6 times; Training a ConvNet (Weight Init, Hyperparameters) – 5 times
  • Unit 4 (Optimization & Generalization): Optimization in Deep Learning – 10 times; Recurrent Networks & LSTM – 6 times; Generalization in Neural Networks – 3 times
  • Unit 5 (Case Studies & Applications): Audio WaveNet – 5 times; Image Classification & Detection – 4 times; NLP & Word2Vec – 3 times