| S.No |
Topic |
Sub-Topics |
| 1 | Keras | Overview, Installation, Backend (TensorFlow), High-level API, Use cases |
| 2 | Core Concepts | Layers, Models, Tensors, Activation functions, Input/Output shapes |
| 3 | Sequential API Basics | Sequential model, Adding layers, compile(), fit(), evaluate() |
| 4 | Sequential API Advanced | Multiple layers, Dropout, Flatten, Dense, Activation layers |
| 5 | Functional API Basics | Input(), Model(), multi-input/output, shared layers, layer connections |
| 6 | Functional API Advanced | Residual connections, Branching models, Merging layers, Concatenate, Add layers |
| 7 | Loss Functions | Binary crossentropy, Categorical crossentropy, MSE, Custom loss, Sparse loss |
| 8 | Optimizers | SGD, Adam, RMSProp, learning rate, custom optimizer |
| 9 | Metrics | Accuracy, Precision, Recall, F1-score, Custom metrics |
| 10 | Callbacks | EarlyStopping, ModelCheckpoint, ReduceLROnPlateau, TensorBoard, custom callbacks |
| 11 | Data Preprocessing | Normalization, Standardization, One-hot encoding, ImageDataGenerator, Tokenizer |
| 12 | Regularization | Dropout, L1/L2, BatchNormalization, Weight constraints, Data augmentation |
| 13 | Dense Networks | Feedforward networks, Layer configuration, Activation functions, Compile, Fit |
| 14 | Convolutional Networks (CNN) | Conv2D, MaxPooling2D, Flatten, Filters, Kernels |
| 15 | Advanced CNN | Transfer learning, Pretrained models, Fine-tuning, Data augmentation, Regularization |
| 16 | Recurrent Networks (RNN) | SimpleRNN, LSTM, GRU, sequence input, return sequences |
| 17 | Advanced RNN | Bidirectional RNN, stacked RNN, attention mechanism, masking, stateful RNN |
| 18 | Embedding Layers | Word embeddings, Embedding layer, pretrained embeddings, input_dim, output_dim |
| 19 | Autoencoders | Encoder-Decoder, Dimensionality reduction, Reconstruction loss, Activation, Custom models |
| 20 | GANs Basics | Generator, Discriminator, Adversarial loss, Training loop, Applications |
| 21 | GANs Advanced | Conditional GAN, DCGAN, WGAN, Training stability, Image generation |
| 22 | Custom Layers | Layer subclassing, call(), build(), trainable weights, forward pass |
| 23 | Custom Models | Model subclassing, forward pass, training loop, GradientTape, metrics |
| 24 | Model Saving & Loading | save(), load_model(), SavedModel format, HDF5 format, weights only |
| 25 | TensorBoard | Visualization, Scalars, Images, Graphs, Embeddings |
| 26 | Performance Optimization | Mixed precision, GPU/TPU usage, Batch size, Learning rate tuning, Profiling |
| 27 | Distributed Training | MirroredStrategy, Multi-GPU, TPU, ParameterServerStrategy, Dataset sharding |
| 28 | Integration with TensorFlow | tf.data, tf.keras, TensorFlow Hub, Custom training loops, SavedModel export |
| 29 | End-to-End NLP Project | Text preprocessing, Tokenization, Embeddings, LSTM/Transformer model, Evaluation |
| 30 | End-to-End CV Project | Data preprocessing, CNN/Transfer learning, Training, Evaluation, Deployment |
