On Tue, Jan 12, 2016 at 9:54 AM, LAU <> wrote: > But, the definition of "deep learning" isn't it "*deep neural network > *learning" > ? The word "deep" doesn't mean "extensive learning" or "learn with much > effort" (*sorry, don't know the exact term, I'm french speaking people*). > The word means "network with many layers". So, "deep learning" is necessary > neural network. > > LAU >
I guess you are right. My recollection is that they not only have multiple layers but that the multiple layers can be specifically defined to respond to different kinds of characteristics that the data may possess and these different processing paths may be recombined in some way. So I think they are like hybrids - at least to some extent. But even if Deep Learning refers to Neural Networks I still believe that some similar characteristics can be found in other kinds of hybrids. However, I may have been projecting my personal views onto the consensus definition a little too vigorously. If you look at the description in Wikiepedia it seems, for example, to refer to non-linear transformations which are not necessarily neural networks. My view is that there are other kinds of networks which can be processed in ways to >From Wikipedia: *Deep learning* (*deep structured learning*, or *hierarchical learning*, or sometimes *DL*, or more correctly *deep machine learning*) is a branch of machine learning <https://en.wikipedia.org/wiki/Machine_learning> based on a set of algorithms <https://en.wikipedia.org/wiki/Algorithm> that attempt to model high-level abstractions in data by using multiple processing layers with complex structures, or otherwise composed of multiple non-linear transformations <https://en.wikipedia.org/wiki/Linear_transformation>. Various deep learning architectures such as deep neural networks <https://en.wikipedia.org/wiki/Deep_learning#Deep_neural_networks>, convolutional deep neural networks <https://en.wikipedia.org/wiki/Convolutional_neural_network>, deep belief networks <https://en.wikipedia.org/wiki/Deep_belief_network> and recurrent neural networks <https://en.wikipedia.org/wiki/Recurrent_neural_network> have been applied to fields like computer vision <https://en.wikipedia.org/wiki/Computer_vision>, automatic speech recognition <https://en.wikipedia.org/wiki/Automatic_speech_recognition>, natural language processing <https://en.wikipedia.org/wiki/Natural_language_processing>, audio recognition and bioinformatics <https://en.wikipedia.org/wiki/Bioinformatics> where they have been shown to produce state-of-the-art results on various tasks. Alternatively, *deep learning* has been characterized as a buzzword <https://en.wikipedia.org/wiki/Buzzword>, or a rebranding of neural networks <https://en.wikipedia.org/wiki/Neural_network> ------------------------------------------- AGI Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/21088071-f452e424 Modify Your Subscription: https://www.listbox.com/member/?member_id=21088071&id_secret=21088071-58d57657 Powered by Listbox: http://www.listbox.com
