AlexNet and the deep learning revolution
AlexNet y la revolución del aprendizaje profundo
- Date
- 2012 · confidence high
- Place
- University of Toronto, Canada
- Credited to
- Alex Krizhevsky, Ilya Sutskever and Geoffrey Hinton; the ImageNet database by Fei-Fei Li (Stanford, 2009)
- People
- 👤 Alexander Krizhevsky2000–👤 Geoffrey Hinton1947–👤 Ilya Sutskever1986–
- Field
- Knowledge & symbols
What it is
Deep convolutional network trained on two GPUs that halved the error rate in the ImageNet competition. It convinced the industry that deep learning worked and set off the AI race.
What it led to
- The mass adoption of deep learning by Google, Facebook and Baidu
- The rise of Nvidia and demand for GPUs
- Computer vision in phones, cars and surveillance
Built on
- Backpropagation for training neural networks1986
- Convolutional neural networks (LeNet)1989-1998
- Graphics processing unit (GeForce 256)1999
- CUDA general-purpose GPU computing2007
- LeNet convolutional networks
- GPU and CUDA
- ImageNet
Made possible
- Connectome1986-2024
- Self-driving car (VaMoRs, DARPA Grand Challenge, Waymo)1986-2018
- Voice assistant (Siri) and smart speaker (Amazon Echo)2011-2014
- Generative adversarial networks (GANs)2014
- AlphaGo2016
- Transformer architecture2017
- AI protein structure prediction (AlphaFold)2020-2021
- Implanted brain-computer interface (BrainGate, Neuralink)2004-2024
- facial recognition