Chapter 8: Deep Learning with TensorFlow/Keras

🛠️ This chapter teaches you how to build, train, and deploy Deep Learning models using TensorFlow and Keras — two of the most popular deep learning frameworks.

---

🔹 1. What Is TensorFlow and Keras?

Tool	Description
TensorFlow	Open-source ML framework by Google, used for building & deploying deep learning models
Keras	High-level API running on top of TensorFlow — simple, fast, and user-friendly

👉 TensorFlow = Engine
👉 Keras = Interface (Easy to use)

---

🔹 2. Steps to Build a Deep Learning Model

plaintext
1. Import Libraries
2. Load and Preprocess Data
3. Build the Model
4. Compile the Model
5. Train the Model
6. Evaluate & Predict

---

🔸 Example: Binary Classifier with Keras

python
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Dense

# 1. Build model
model = Sequential()
model.add(Dense(32, input_dim=10, activation='relu'))
model.add(Dense(1, activation='sigmoid'))

# 2. Compile model
model.compile(optimizer='adam', loss='binary_crossentropy', metrics=['accuracy'])

# 3. Train model
model.fit(X_train, y_train, epochs=10, batch_size=32)

---

🔹 3. CNNs (Convolutional Neural Networks) — for Image Data

CNNs are best for image classification, object detection, and computer vision.

CNN Layers:

Layer	Function
Conv2D	Extract features (edges, textures)
MaxPooling2D	Downsamples the image
Flatten	Converts to 1D vector
Dense	Fully connected layer

python
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Conv2D, MaxPooling2D, Flatten, Dense

model = Sequential([
    Conv2D(32, (3,3), activation='relu', input_shape=(28,28,1)),
    MaxPooling2D(2,2),
    Flatten(),
    Dense(64, activation='relu'),
    Dense(10, activation='softmax')
])

📌 Use Case: Handwritten digit recognition (MNIST)

---

🔹 4. RNNs (Recurrent Neural Networks) — for Sequential Data

RNNs remember previous inputs and are used in text, speech, and time-series.

Common RNN Types:

Type	Use
Simple RNN	Basic sequence modeling
LSTM	Long-Term memory — handles long dependencies
GRU	Faster version of LSTM

python
from tensorflow.keras.layers import LSTM, Embedding

model = Sequential()
model.add(Embedding(input_dim=5000, output_dim=64))
model.add(LSTM(128))
model.add(Dense(1, activation='sigmoid'))

📌 Use Case: Sentiment analysis, text classification

---

🔹 5. Hands-On Projects

Project	Tools & Use
Image Classification	CNN + MNIST dataset
Sentiment Analysis	LSTM + IMDB reviews
Object Detection	TensorFlow Object Detection API
Chatbot	RNN + NLP preprocessing

---

🔹 6. Model Saving & Deployment

python
# Save model
model.save('model.h5')

# Load model
from tensorflow.keras.models import load_model
model = load_model('model.h5')

For deployment:

• Use Flask or FastAPI

• Export model as .h5 or .tflite for mobile

---

🧠 Summary of Chapter 8

Concept	Details
TensorFlow	Core DL framework by Google
Keras	Easy API for building models
CNN	Works best with image data
RNN/LSTM	Works best with sequence data
Projects	Classification, sentiment, object detection
Tools	Google Colab, Jupyter, TensorBoard

---

✅ Mini Assignments

1. Build a CNN to classify handwritten digits (MNIST).

2. Train an LSTM model to classify movie reviews.

3. Save and reload a trained Keras model.