Deep learning is a Connectionism approach and is one type of neural networks.
In the 1970s, with regards to the combination of neurons in brain's visual cortex, Torsten Nils Wiesel and David Hunter Hubel's model was known. *
In 2006, Auto encoder, a new method of learning neural networks was discovered, and it began to attract attention from around 2010.
Unlike the conventional neural networks, it can identify the characteristic quantities of input images and waveforms without humans having to look for them. Although neural networks have been deepened by Auto encoder to have multiple layers, the essential change is the automatic extraction of characteristic quantities.
■References (Books, papers, Web articles, etc.)
「ネオコグニトロンはまだ進化する」、画像向けディープラーニング「CNN」の父に聞く 日経テクノロジーオンライン http://techon.nikkeibp.co.jp/article/INTERVIEW/20150521/419523/
*: This work was awarded the Nobel Prize in Physiology or Medicine in 1981. Kunihiko Fukushima at NHK, with this work as a hint, made the neural networks Neocognitron (1979), which was able to recognize various forms of letters. What was used here is a convolutional neural network (CNN), which later became the basis for deep learning.
★If you interested in this text, please have the following ebook at hand. For that, Click on the URL below.
https://society-zero.com/icardbook/010/index.html
★This article is a piece of "knowledge card" that makes up iCardbook, IAI meets Philosophy: how to design the Game AI (Yoichiro Miyake).
iCardbook, a card-type specialized book that organizes “Knowledge Cards” is a new form of book in the smartphone age.
■iCardbook Benefits:Good value for reading
・you can read it quickly in your pocket of time, e.g. commuting
・You can immediately understand the "drawing " in the author's head
・Amazon Prime members can read for free
◎About iCardbook(english)
https://society-zero.com/icard/icardbookenglish
◎About iCardbook(japanese)
https://society-zero.com/chienotane/archives/5067
■References list of "AI meets Philosophy: how to design the Game AI"
