Machine learning is a scientific field in which instead of trying to code a limied behaviour program, we try to develop an algorithm that a computer can use to look at hundreds or thousands of examples (and the correct answers), and then the computer uses that experience to solve the same problem in new situations. Essentially, our goal is to teach the computer to solve by example, very similar to how we might teach a young child to distinguish a cat from a dog.

One of the big challenges with traditional machine learning models is a process called feature extraction. Specifically, the programmer needs to tell the computer what kinds of things it should be looking for that will be informative in making a decision. So it’s like teaching a computer over and again with a cat sitting on table or cat sitting in a garden. We never have to do that to a human brain, a child once taught what cat looks like they automatically points out a cat no matter where in the image cat is irrespective of whether cat is sleeping or playing or if the cat is white domestic or a black wild one. That is the whole idea behind Deep learning, that is to apply principles of human brain.

Feeding the algorithm raw data rarely ever works, so feature extraction is a critical part of the traditional machine learning workflow. This places a huge burden on the programmer, and the algorithm's effectiveness relies heavily on how insightful the programmer is. For complex problems such as object recognition or handwriting recognition, this is a huge challenge.

Deep Learning involves training a Neural Network like structures with 2 or more hidden layers also sometimes called Deep neural networks. The more big data set to learn, the more intelligent and accurate the network becomes.

Deep learning works on the principles of human brain. It is one of the only methods by which we can circumvent the challenges of feature extraction. This is because deep learning models are inspired by the functionality of human brain and are capable of learning to focus on the right features by themselves, requiring little guidance from the programmer. This makes deep learning an extremely powerful tool for modern machine learning.

Deep learning networks can work in both Supervised (Classification) and Unsupervised (Pattern analysis) manner. They can be trained by providing sample inputs/outputs to autocorrect or they can solely learn on their own over time to yield more accurate output on their own.

To some extent but not entirely. Infact in some use cases both combined may yield a highly intelligent solution.

Given two models that perform more or less equally, you should always prefer the one that is less complex. For this reason, there will always be cases where Deep Learning will not be preferred, even if it has managed to squeeze an extra 1% in accuracy on the testing set.

Deep Learning is more suited to highly complicated situations that no programmer can pre-train their algorithms on with 100% accuracy like Image/Voice recognition, Natural language processing to name a few. These situations demands computers to learn and infer on their own and become accurate with experience like our brain does.

For many applications, far simpler traditional algorithms like logistic regression or support vector machine will work just fine, and using a deep belief network will only complicate things.

Deep Learning is very close to solving supervised learning in the asymptote of training data size, and that's going to push some traditional learning algorithms to near extinction.

Handwriting recognition and interpretation: In fact, the best commercial neural networks are now so good that they are used by banks to process cheques, and by post offices to recognize addresses.

Real time Fraud decision: PayPal engineering blogs, we learn that they are using deep learning via Hadoop to build intelligence into the network. Banks, payment processors and other financial firms will soon move to real-time analytics and artificial intelligence techniques to crack down on fraud.

Detecting cancer: A new feature in Samsung Medison's ultrasound system uses a deep-learning algorithm to make recommendations about whether a breast abnormality is benign or cancerous. this avoids unnecessary biopsies.

Natural Language Processing which is used heavily in language conversion in chat rooms or processing text from human speeches.

Image & Optical Character Recognition which is scanning of images. It's gaining traction lately to read an image and extract text out of it and correlate to the objects found on image. Google uses his technique to list images based on keyword provided.

Speech Recognition applications like Apple Siri or Microsoft Cortana needs no introduction

Drug discovery though medical imaging-based diagnosis using deep learning. It's kind of in early stages now. Please refer to the Butterfly Network for the work they are doing.

CRM needs for companies are growing day by day. Every company wants to know their potential customers. Deep Learning has provided some outstanding results.