💡 Keras: Building Neural Networks Simply

Keras is a high-level deep learning API, now part of TensorFlow, designed for fast and easy experimentation. This guide covers the fundamental workflow: defining, compiling, training, and using a neural network model.

from tensorflow import keras
from tensorflow.keras import layers

# Define a Sequential model
model = keras.Sequential([
    # Input layer with 64 neurons, expecting flat input data
    layers.Dense(64, activation="relu", input_shape=(784,)),
    # A hidden layer with 32 neurons
    layers.Dense(32, activation="relu"),
    # Output layer with 10 neurons for 10-class classification
    layers.Dense(10, activation="softmax")
])

model.summary()

• Model Definition: keras.Sequential creates a simple, layer-by-layer model.
• layers.Dense is a standard fully-connected layer. The first layer must specify the input_shape.
• activation functions like "relu" introduce non-linearity, while "softmax" is used on the output layer for multi-class classification to produce probabilities.

# (Continuing from the previous step)
model.compile(
    optimizer='adam',
    loss='sparse_categorical_crossentropy',
    metrics=['accuracy']
)

print("Model compiled successfully.")

• Compilation: .compile() configures the model for training.
• optimizer is the algorithm used to update the model's weights (e.g., 'adam' is a popular choice).
• loss is the function the model tries to minimize during training. sparse_categorical_crossentropy is common for integer-based classification labels.
• metrics are used to monitor the training and testing steps. Here, we track accuracy.

import numpy as np

# Create dummy training data
x_train = np.random.random((1000, 784))
y_train = np.random.randint(10, size=(1000,))

# Train the model
history = model.fit(
    x_train,
    y_train,
    epochs=5,
    batch_size=32,
    verbose=0 # Hides the progress bar for a cleaner output
)

print(f"Training complete. Final accuracy: {history.history['accuracy'][-1]:.4f}")
# Output (will vary):
# Training complete. Final accuracy: 0.4570

• Training: The .fit() method trains the model on your data.
• x_train and y_train are your input features and target labels.
• epochs defines how many times the model will see the entire dataset.
• batch_size is the number of samples processed before the model is updated.

# Create a single dummy sample to test
x_test = np.random.random((1, 784))

# Get the model's prediction
predictions = model.predict(x_test)
predicted_class = np.argmax(predictions[0])

print(f"Predicted class: {predicted_class}")
print(f"Confidence scores: {predictions[0].round(2)}")
# Output (will vary):
# Predicted class: 3
# Confidence scores: [0.09 0.1  0.1  0.12 0.1  0.09 0.11 0.1  0.09 0.1 ]

• Prediction: .predict() is used to make predictions on new, unseen data.
• For a classification model with a softmax output, this returns an array of probabilities for each class.
• np.argmax() is used to find the index (the class) with the highest probability score.

#Keras #TensorFlow #DeepLearning #MachineLearning #Python

━━━━━━━━━━━━━━━
By: @CodeProgrammer ✨

🤖🧠 The Transformer Architecture: How Attention Revolutionized Deep Learning

🗓️ 11 Nov 2025
📚 AI News & Trends

The field of artificial intelligence has witnessed a remarkable evolution and at the heart of this transformation lies the Transformer architecture. Introduced by Vaswani et al. in 2017, the paper “Attention Is All You Need” redefined the foundations of natural language processing (NLP) and sequence modeling. Unlike its predecessors – recurrent and convolutional neural networks, ...

#TransformerArchitecture #AttentionMechanism #DeepLearning #NaturalLanguageProcessing #NLP #AIResearch