The Forward-Forward Algorithm: Some Preliminary Investigations
By Geoffrey E. Hinton. arXiv.org, 2022
443 citations recorded by the source corpus.
The aim of this paper is to introduce a new learning procedure for neural networks and to demonstrate that it works well enough on a few small problems to be worth further investigation. The Forward-Forward algorithm replaces the forward and backward passes of backpropagation by two forward passes, one with positive (i.e. real) data and the other with negative data which could be generated by the network itself. Each layer has its own objective function which is simply to have high goodness for positive data and low goodness for negative data. The sum of the squared activities in a layer can be used as the goodness but there are many other possibilities, including minus the sum of the squared activities. If the positive and negative passes could be separated in time, the negative passes could be done offline, which would make the learning much simpler in the positive pass and allow video to be pipelined through the network without ever storing activities or stopping to propagate derivatives.
Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition
View Paper: https://www.semanticscholar.org/paper/75e3475cf49caf1dbbcad526b0132b455dc88dd5
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Artificial Intelligence
- Sanjeev Arora
- Jason D. Lee
- Danqi Chen
- Tianyu Gao
- Tengyu Ma
- Luke S. Zettlemoyer
- Sadhika Malladi
- Percy Liang
In Computer Vision and Pattern Recognition
- Jason D. Lee
- Danqi Chen
- Tianyu Gao
- Miao Hu
- Sanjeev Arora
- Eshaan Nichani
- Alexandru Damian
- Sadhika Malladi
Citations in the corpus
- Fine-Tuning Language Models with Just Forward Passes (2023, methodology): “Additionally, there are ZO methods that optimize without estimating the gradient [36, 66, 42].”
- Training Spiking Neural Networks Using Lessons From Deep Learning (2021, unknown): “Initially proposed by Dellaferrera and Kreiman [145], Hinton’s [146] forward– forward learning algorithm generated more traction soon after.”
- Thousands of conductance levels in memristors integrated on CMOS (2023, background): “Moreover, such analog memristors may also enable electronic circuits capable of growing for the recently proposed mortal computations 30 .”