Deep learning; The network of Artificial Neurons Learn through Learning to achieve excellence “
Deep learning is an aspect of artificial intelligence inspired from the structural and functional behavior of brain(the network of billion’s of the neuron), operates with an approach to learn things the similar way human brain did.
“Simply say learning from experience and observation”
Even there were so many definitions of deep learning(DL) out there. Consequently, the main approach that almost every one explaining its multi-linked artificial neurons structure design to mimic the brain. And learn progressively and get better and better through time and experience.
Deep Learning vs Machine learning
While traditional machine learning algorithms are linear, DL algorithms are stacked in a hierarchy of increasing complexity and abstraction. Sometimes people call them the “black box” of learning. To understand deep learning. For instance, imagine a toddler who just learns how to toddle. Has no idea how to differentiate between hot and cold, The parents told him don’t touch it it’s hot it harm you. But the curiosity always brings child near to danger and once toddler experience the warmth. Got the lesson need not touch something which is hot.
Through time he becomes more aware of the features of hotness. As a result, what toddler did, building a hierarchy in which each level of abstraction is created with the knowledge that gained from Experience. That’s how the neural network learns.
In traditional machine learning, the learning process is supervised and the programmer has to be very very specific when telling the computer what types of things it should be looking for when deciding if an image contains a dog or cat. Long tung, long ear, bigger eye, long tail, should be labeled “dog.”
This is a laborious process called feature extraction. For which programmer have to bog down into detail and find the correct the set of features, and success rate depends entirely upon the programmer’s ability to accurately define a feature set for “dog.”
But in DL programmer, don’t need to get down into the details of feature extraction or feature learning. AI Program builds the feature set by itself without supervision. The program uses the information it receives in the form of training data to create a feature set for dog and build a predictive model. Through data and learning. The feature building process for predictive model become more and more accurate. More data Better set of feature.
Deep learning — are often called:- “Data hungry” means it takes lots of data to make the system work.
What experts and leader said of deep learning.
“Deep Learning is Large Neural Networks”
He has spoken and written a lot about what deep learning is and is a good place to start.
he described the idea of deep learning as:
Using brain simulations, hope to:
– Make learning algorithms much better and easier to use.
– Make revolutionary advances in machine learning and AI.
I believe this is our best shot at progress towards real AI
He provides a nice cartoon of this in his slides: which explains how the performance of deep learning models gets better and better when data grows gradually.
Yoshua Bengio is another leader in deep learning although began with a strong interest in the automatic feature learning that large neural networks are capable of achieving.
He describes deep learning in terms of the algorithms ability to discover and learn good representations using feature learning. In his 2012 paper titled “DL learning of Representations for Unsupervised and Transfer Learning”
“Deep learning algorithms seek to exploit the unknown structure in the input distribution in order to discover good representations, often at multiple levels, with higher-level learned features defined in terms of lower-level features”.
In a 2016 talk he gave titled “DL and Understandability versus Software Engineering and Verification” he defined deep learning in a very similar way to Yoshua, focusing on the power of abstraction permitted by using a deeper network structure.