Functional API basics in TensorFlow - Model Pipeline Trace

Model Pipeline - Functional API basics

This pipeline shows how to build a simple neural network using TensorFlow's Functional API. It takes input data, processes it through layers, trains the model, and then makes predictions.

Data Flow - 4 Stages

1Input Layer

1000 rows x 10 features→Define input shape for the model→1000 rows x 10 features

[[0.5, 1.2, ..., 0.3], [0.1, 0.4, ..., 0.7], ...]

↓

2Dense Layer 1

1000 rows x 10 features→Apply fully connected layer with 8 neurons and ReLU activation→1000 rows x 8 features

[[0.0, 1.5, ..., 0.3], [0.2, 0.7, ..., 0.1], ...]

↓

3Dense Layer 2

1000 rows x 8 features→Apply fully connected layer with 4 neurons and ReLU activation→1000 rows x 4 features

[[0.1, 0.4, 0.0, 0.2], [0.3, 0.1, 0.5, 0.0], ...]

↓

4Output Layer

1000 rows x 4 features→Apply output layer with 1 neuron and sigmoid activation for binary classification→1000 rows x 1 feature

[[0.7], [0.2], [0.9], ...]

Training Trace - Epoch by Epoch

Loss
0.7 |****
0.6 |*** 
0.5 |**  
0.4 |*   
0.3 |*   
     1 2 3 4 5 Epochs

Epoch	Loss ↓	Accuracy ↑	Observation
1	0.65	0.60	Model starts learning, loss is high, accuracy is low
2	0.50	0.75	Loss decreases, accuracy improves
3	0.40	0.82	Model continues to learn well
4	0.35	0.86	Loss decreases further, accuracy increases
5	0.30	0.89	Training converges with good accuracy

Prediction Trace - 4 Layers

Layer 1: Input Layer

Layer 2: Dense Layer 1 (ReLU)

Layer 3: Dense Layer 2 (ReLU)

Layer 4: Output Layer (Sigmoid)

Model Quiz - 3 Questions

What does the output layer's sigmoid activation do?

A Sets all negative values to zero

B Converts values to a probability between 0 and 1

C Normalizes data to have zero mean

D Reduces dimensionality of input

Key Insight

Using the Functional API lets us clearly define how data flows through each layer. We see the model learns by reducing loss and improving accuracy, and the output layer produces probabilities for classification.