Paper calendar

Daily

One paper a day, summarised and placed in context. Ten focused minutes with a question to carry into the reading; the full order of 36 papers comes around again every 36 days.

Today · Monday, Sep 14, 2026 · Day 45

Gradient-Based Learning Applied to Document Recognition

ML math · LeCun, Bottou, Bengio, Haffner · 1998 · Proceedings of the IEEE · 70 min · Moderate

The paper that put convolutional networks on a firm footing: local receptive fields, shared weights, and pooling give translation-tolerant features with far fewer parameters than dense layers. It also describes training whole systems end to end with gradients, including document-level pipelines.

Question to keep in mind

How many parameters does a convolutional layer have compared with a dense layer over the same input, and why does that gap matter for generalisation?

Day45
September 2026
MonTueWedThuFriSatSun123456789101112131415161718192021222324252627282930
AI researchSystemsML math

Sign in to mark briefs as read and keep a streak.

Recent briefs
Day 45 · Sep 14Gradient-Based Learning Applied to Document RecognitionOpen
Day 44 · Sep 13The Google File SystemOpen
Day 43 · Sep 12Deep Residual Learning for Image RecognitionOpen
Day 42 · Sep 11Learning representations by back-propagating errorsOpen
Day 41 · Sep 10The Log-Structured Merge-Tree (LSM-Tree)Open
Day 40 · Sep 9Adam: A Method for Stochastic OptimizationOpen
Day 39 · Sep 8A Mathematical Theory of CommunicationOpen
Day 38 · Sep 7Time, Clocks, and the Ordering of Events in a Distributed SystemOpen
Day 37 · Sep 6ImageNet Classification with Deep Convolutional Neural NetworksOpen
Day 36 · Sep 5Denoising Diffusion Probabilistic ModelsOpen
Day 35 · Sep 4Kafka: a Distributed Messaging System for Log ProcessingOpen
Day 34 · Sep 3Direct Preference Optimization: Your Language Model is Secretly a Reward ModelOpen
Day 33 · Sep 2Deep Double Descent: Where Bigger Models and More Data HurtOpen
Day 32 · Sep 1Spanner: Google's Globally-Distributed DatabaseOpen
124 online