References: Deep learning
History:
In 2006, Hinton gave a breakthrough step.
In 2013, Robin Li announced the establishment of the Deep Learning Institute.
At present, many fields of State-of-the-art are occupied by deep learning.
Significance:
Typically, traditional methods require manual extraction of features:
Although, there are many good features, such as SIFT, hog features, but manual extraction features are very not, and not enough "mechine learning". Also, for a given question, what characteristics do you want to choose? Would you try it one by one or more and then summarize a set of experiences? Obviously, this kind of "mechine learning" makes people suspect. So
Deep learning came out.
Two classic examples:
CNN (supervised learning)
RBN (unsupervised learning)
Why so good things, the future so wonderful things, I still hesitate so long not to learn? Too conservative!!! This kind of rigid thought!! My bro every day to urge me to learn, I am too happy. Fight It!
Biological Bionic
Its thought comes from the connection of neurons in the brain. This is OK, the creature's things are always elusive.
In addition, the feature extraction of this analogy is used.
Deep learning--history and significance