Gradient-Based Learning Applied to Document Recognition

Authors: LeCun, Y., Bottou, L., Bengio, Y., & Haffner, P. Year: 1998 Venue: Proceedings of the IEEE

Summary

LeNet introduces the first practical CNN applied to handwritten digit recognition. It demonstrates the power of convolutional layers, pooling, and backpropagation for visual tasks.

Key Concepts

• Convolutional layers for feature extraction
• Pooling for dimensionality reduction
• Backpropagation in deep networks
• Efficient gradient computation
• End-to-end trainable architecture

Impact

LeNet laid the groundwork for all modern CNNs. Despite being from 1998, its principles remain fundamental to computer vision today.

Related Papers

• AlexNet: ImageNet Classification
• VGGNet: Very Deep Convolutional Networks

---

Citation:

@article{lecun1998gradient,
  title={Gradient-based learning applied to document recognition},
  author={LeCun, Yann and Bottou, L{\'e}on and Bengio, Yoshua and Haffner, Patrick},
  journal={Proceedings of the IEEE},
  volume={86},
  number={11},
  pages={2278--2324},
  year={1998}
}