#3 Neural networks

The Voice recognition and processing is a tough task. Google has now shifted to using a neural network that’s much more effective in speech recognition. A neural network is a computer system that imitates neurons of the brain. It is layers of software-based “neurons” that work like actual neurons, like take input impulse in and fire out put impulse off to other neurons based on the data received.

The results of research led by Google Fellow Jeff Dean's on neural networks made some waves. Google had taught a computer to recognize cats in videos. The catch is that neural network created the concept of "cat" on its own without direct human intervention.

The first layer of neuron software looks for simple data, like angled lines or colors. If it matches what it is looking for, it simply fires off a signal. And then second layer of neurons pays attention to signal from first layer. As more and more neuron layers gets added to create conceptual abstraction until at the very top layer a neuron gets trained to recognize cats 15.8 percent of the time.