Neural Networks Need Naps, Just Like You

19 36

A neural network works and is modelled similarly to the human brain’s neural network. It is a specialized branch of machine learning.

It's training process is in a manner analogous to how humans and other biological systems learn from their environment during childhood development.

While powering many systems, filtering email spam, and even aiding in cancer research, neural network deserves a bit of rest and relaxation. In fact, it would like a day off.

According to new research from the Los Alamos National Laboratory, neural network might actually need that shuteye(sleep)—or at least a machine's version of it.

Because neural net simulations can become unstable after prolonged periods of unsupervised learning, a digital kind of "nap" could return the system to stasis.

So, to achieve some semblance of artificial sleep, the researchers had to inject noise (Gaussian noise waves) into the neural network. It is equivalent to a good night rest for it. Thus the network again becomes stable again and starts working.

Special thanks to @Ashma for this great support and cordial help to everyone.

If you want to be sponsored, please join the following community:

https://read.cash/c/get-sponsored-2a0b

You can not only get sponsorship here but also can earn points by doing some simple tasks. Thanks again @Ashma for creating this amazing community.

7
$ 0.00
Sponsors of Tamanna21
empty
empty
empty

Comments

Nice article. Keep doing your best.

$ 0.00
4 years ago

thank you for the support

$ 0.00
4 years ago

Nice article.

$ 0.00
4 years ago

thanks dear

$ 0.00
4 years ago

Welcome

$ 0.00
4 years ago

This article is very important. Your write way is very good. Keep it up dear, Best wishes for you💕

$ 0.00
4 years ago

thank you dear for supporting me

$ 0.00
4 years ago

Amazing article👏👏👏🤩

$ 0.00
4 years ago

haha thanks dear

$ 0.00
4 years ago

Great article.. Keep writing ❤

$ 0.00
4 years ago

thanks dear...getting support from you inspires me to write more

$ 0.00
4 years ago