Applause from Ludovic Benistant, Mate Labs, and 39 others
Kailash Ahirwar
Co-founder @ Mate Labs | Democratizing Arti cial Intelligence
Jan 30 · 2 min read
Why we need a better learning algorithm
than Backpropagation in Deep Learning
We all agree on one thing that Back propagation is a revolutionary
learning algorithm. For sure, it has helped us in the training of almost
all neural network architectures. With the help of GPUs,
backpropagation has reduced months of training time to hours/days of
training time. It has allowed an e cient training of neural networks.
I think of two reasons because of which it has gotten this widespread
adoption, (1) we didn’t have anything better than backpropagation, &
(2) it worked. Backpropagation is based on the chain rule of
di erentiation.
Source — toptal -https://www.toptal.com/
Problem lies in the implementation of the Backpropagation algorithm
itself. To calculate gradients of the current layer we need gradients of
the next layer
raining/Backpropagation/backpropagation/time/neural/net/training/thing/algorithm/机器/
raining/Backpropagation/backpropagation/time/neural/net/training/thing/algorithm/机器/
-->