Skip to content

Gradient-Based Learning Applied to Document Recognition (LeNet)

Authors: Yann LeCun, Léon Bottou, Yoshua Bengio, Patrick Haffner Year: 1998 Citations: 30,000+

Summary

Seminal CNN work for document recognition via LeNet-5. Demonstrates effectiveness of convolutional layers with weight sharing. Foundation for modern computer vision deep learning.

Key Concepts

  • Convolutional Layers: Local receptive fields with weight sharing
  • Local Feature Extraction: Filters detect edges, textures, patterns
  • Pooling Layers: Downsampling while preserving information
  • Weight Sharing: Parameter efficiency via filter reuse
  • Hierarchical Learning: Simple to complex features

Impact

  • 30,000+ citations
  • Foundational for computer vision deep learning
  • Demonstrated CNNs work for real tasks
  • Influenced AlexNet, ResNet, VGGNet, etc.
  • Essential foundation for visual learning
  • Decades of practical deployment
  • AlexNet (Krizhevsky et al., 2012)
  • VGGNet (Simonyan & Zisserman, 2014)
  • ResNet (He et al., 2015)