cnn

Convolutional Neural Network

Architecture: Input → convolution → pooling → Dense layers → Output

Extracts localized features from images

Handles pixel dependencies

Unaffected by positional change (translation invariance)

Uses relu activation

Filters found using gradient-descent

Vulnerable to adversarial-attack

---

*References

• CNN Explained - Stanford CS231n
• Convolutional Neural Networks - TensorFlow #ml-notes