By Qiaqia Li
One solution is to allow computers to learn from experience and understand the word in terms of ** a hierarchy of concepts**, with each concept defined through its relation to simpler concepts. By gathering knowledge from experience, this approach avoid the need for human operators to formally specify all the knowledge that the computer needs. The hierarchy of concepts enables the computer to learn complicated concepts by building them out of simpler ones. If we draw a graph showing how these concepts are build on top of each other, the graph is deep, with many layers. For this reason, we call this approach to AI deep learning.
The capability of machine to acquire their own knowledge by extracting patterns from raw data.
A simple machine learning algorithm can solve binary problem, like recognizing whether there is a cat in the picture, or determining whether to recommend cesarean delivery.
Each piece of information included in the representation of the input data.
One solution is to use machine learning to discover not only the mapping from the representation to output but also the representation itself.
The quintessential example of a representation learning algorithm. An autoencoder is the combination of an encoder function, which converts the input data into a different representation, and a decoder function, which converts the new representation back into the original format. Autoencoders are trained to preserve as much information as possible when an input is run through the encoder and then the decoder, but they are also trained to make the new representation have various nice properties. Different kinds of antoencoders aim to achieve different kinds of properties.
By the way, Geoffrey Hinton recently published a new paper called Stacked Capsule Autoencoders , here is **the interpretation article **written by his fellow.