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
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
AI researchSystemsML math
Sign in to mark briefs as read and keep a streak.
Gradient-Based Learning Applied to Document RecognitionOpen
The Google File SystemOpen
Deep Residual Learning for Image RecognitionOpen
Learning representations by back-propagating errorsOpen
The Log-Structured Merge-Tree (LSM-Tree)Open
Adam: A Method for Stochastic OptimizationOpen
A Mathematical Theory of CommunicationOpen
Time, Clocks, and the Ordering of Events in a Distributed SystemOpen
ImageNet Classification with Deep Convolutional Neural NetworksOpen
Denoising Diffusion Probabilistic ModelsOpen
Kafka: a Distributed Messaging System for Log ProcessingOpen
Direct Preference Optimization: Your Language Model is Secretly a Reward ModelOpen
Deep Double Descent: Where Bigger Models and More Data HurtOpen
Spanner: Google's Globally-Distributed DatabaseOpen