Deep Learning

6 milestones in AI history

Geoffrey Hinton, pioneer of backpropagation in neural networks
ResearchExpert Systems Boom

Backpropagation Rediscovered

Rumelhart, Hinton, and Williams published 'Learning Representations by Back-propagating Errors' in Nature, demonstrating that backpropagation could train multi-layer neural networks effectively. The same year, the PDP (Parallel Distributed Processing) group published their influential two-volume work on connectionism.

David RumelhartGeoffrey HintonUC San DiegoCarnegie Mellon University
LeNet-5 convolutional neural network architecture
ResearchSecond AI Winter

LeNet: Convolutional Neural Networks

Yann LeCun demonstrated that convolutional neural networks (CNNs) could be trained with backpropagation to recognize handwritten digits. The refined LeNet-5 (1998) achieved 99%+ accuracy on MNIST and was deployed by banks to read checks — running in ATMs for years.

Yann LeCunAT&T Bell Labs
Deep belief network architecture diagram
ResearchDeep Learning Dawn

Deep Belief Networks: Hinton Revives Deep Learning

Geoffrey Hinton published 'A Fast Learning Algorithm for Deep Belief Nets,' showing that deep neural networks could be effectively trained by pre-training each layer as a restricted Boltzmann machine. This solved the long-standing problem of training networks with many layers.

Geoffrey HintonSimon OsinderoUniversity of Toronto
AlexNet deep neural network architecture diagram
ResearchDeep Learning Breakthrough

AlexNet: The ImageNet Moment

AlexNet, a deep convolutional neural network, won the ImageNet competition by a staggering margin — reducing the error rate from 26% to 16%. Trained on two NVIDIA GTX 580 GPUs, it was dramatically deeper and more powerful than previous entries. The AI community was stunned.

Alex KrizhevskyIlya SutskeverUniversity of Toronto
Google DeepMind logo
ResearchDeep Learning Breakthrough

DeepMind's DQN Masters Atari Games

DeepMind demonstrated a deep reinforcement learning agent (Deep Q-Network) that learned to play Atari 2600 games directly from pixel inputs, achieving superhuman performance on many games with no task-specific engineering. Google acquired DeepMind for ~$500 million shortly after.

Volodymyr MnihDemis HassabisDeepMind
Residual network skip connection block diagram
ResearchDeep Learning Breakthrough

ResNet: Deeper Than Ever

Microsoft Research introduced ResNet with skip connections (residual connections), enabling the training of networks with 152+ layers — 8x deeper than previous networks. ResNet won ImageNet 2015 with 3.57% error, surpassing human-level performance (5.1%) for the first time.

Kaiming HeXiangyu ZhangMicrosoft Research

Related Topics