Artificial Intelligence: Google Brain

Google Brain is an artificial intelligent machine which speaks to the user and answers the questions. It is a deep learning research project at Google. Andrew Ng, the Director of the Stanford Artificial Intelligence Lab founded the Google Brain project at Google, which developed very large scale artificial neural networks using Google's distributed compute infrastructure.

Google Brain is a fascinating project which uses set of algorithms from machine learning. It also uses artificial neural networks. Researchers try to train the Google Brain uses various sets of training data so that it can be used to enable high quality speech recognition, email spam blocking etc. Even the Google’s self-driving car concept is a part of the Google Brain project.

To teach the machine to distinguish between cars and motorcycles, the researchers need to collect tens of thousands of pictures that have already been labeled as ‘car’ or ‘motorcycle’ to train them. People working on this project say that they are create a neural network which simulates a new-born brain and introduce it to YouTube videos for a week. They inferred that it started to learn unlabeled images from the YouTube stills. This small-scale simulation of Google Brain learned what the word ‘cat’ is from the YouTube’s unlabeled stills. This is known as self-taught learning.

They are working on implementing this on all possible areas like speech recognition and email spam blocking.