Deep Learning simplified

-Abbhirami

It has been a very long time since I blogged, I guess.  I was caught up with much stuff. Anyway, today’s topic is one of the coolest topics. Let us not waste time. We’ll dive straight into it.

When I was actually hearing this term for the very first time, I was little confused with its difference with Machine learning, Artificial Neural networks and Artificial intelligence. These terms are kind of having a little similar sense.
Before getting to understand, it is must to know the difference between each term that was mentioned above.

All these terms were given to machines that can simply behave and think more like humans.
Artificial Intelligence is the term given to machine that has the capacity to take decisions all by itself.  It is under this that every other topic exists.

Machine learning, is, to put short making machines learn through putting them into work , ie. they are not programmed explicitly but are actually in a situation experiencing it and analyzing all by itself. And remember this is a type of Artificial intelligence.

Artificial neural networks, stands as a stepping stone to artificial intelligence. It is because of the artificial neural network that a new way of learning is possible in machines. It is done by constructing a simple program framework to respond to problem and receive feedback.

Coming to Deep Learning, it is actually a branch of Machine Learning (ML) that has a set of algorithms that helps them to analyse and sort things accordingly. It travels through layers,  analysing them.

 I read in an article that this obscure term, deep learning has a set of algorithms that came into public limelight when Google researchers fed 10 million random, unlabelled images from YouTube into their experimental Deep Learning System. They then structured the system to recognise the basic elements of picture and how these elements fit together. The system comprising , 16,000 CPUs was able to identify images that shared similar characteristics(such as images of Cats) This canonical experiment showed the potential of Deep Learning algorithms. Deep learning algorithms apply to many areas including Computer Vision, Image recognition, pattern recognition, speech recognition, behaviour recognition etc.

How computers learn?

Learning through layers

 Deep Learning algorithms are modeled on the workings of the Brain. The Brain may be thought of as a massively parallel analog computer which contains about 10^10 simple processors (neurons) – each of which require a few milliseconds to respond to input. To model the workings of the brain, in theory, each neuron could be designed as a small electronic device which has a transfer function similar to a biological neuron. We could then connect each neuron to many other neurons to imitate the workings of the Brain. In practice,  it turns out that this model is not easy to implement and is difficult to train.

It is an approach called bottom –up where in the machines are trained with the inputs which they process with their power processing CPUs in achieving the output which is later used when experiencing  such situation later. One of the important concepts of bottom up approach is representational problem.  “Representations”  are meant to be operated on by algorithms. An algorithm is a systematic way of repeatedly applying mathematical operations to a representation in order to achieve some computational goal. Asking what algorithms a representation supports, therefore, is a matter of asking what mathematical operations can be meaningfully applied to it. 

Representation learning or feature learning, applies algorithms that helps to learn to how to learn.

Before getting into this we must get to know the difference between Supervised and unsupervised learning. In supervised learning, the output data sets are provided which are used to train the machine and get the desired output(labelled data fed) whereas in unsupervised learning no data set is provided, instead the data is clustered into different classes(unlabelled data fed into system).

Deep learning helps achieve a more intuitive way of gathering minute details about the problem and helps to solve them ignoring their complexity. It is an emerging field and over the next few years, due to advances in technology, we are likely to see many more applications in the Deep learning space.

I will post more on this in tomorrow. Give it a thumps up, if you like it