Importance of Deep Learning

Source: Deep Learning on Medium

Importance of Deep Learning

Artificial Intelligence as the name suggests is to make a machine artificially intelligent i.e making the machine think or act like humans. It’s not that the concept of AI is totally new instead it has been explored many decades ago but in recent years it has gained popularity and has lead to various miraculous discoveries because of 2 major factors. The 2 factors which have made all the world invest in this field are increase in computational speed and the amount of useful data available. In fact the 90% of the total data available as of now has been gathered in the previous 3 to 4 years itself .If a robot is hard coded i.e. all the logic has been coded manually into the system then it’s not AI so it can’t be said that simply robots means AI.

To understand AI more clearly let’s dive into its most popular and useful subset, Machine learning. Machine learning simply means making a machine learn from its experience and improving its performance with time just like the case of a human baby. Concept of machine learning became feasible only when sufficient amount of data was made available for training machines. Machine learning helps in dealing with complex and robust systems. But most of the miraculous discoveries which we have come across in recent years have been made possible out of Deep Learning.

Now the question arises what is it in deep learning which has brought such a revolution in our lives . Basically deep learning is itself a subset of machine learning but in this case the machine learns in a way in which humans are supposed to learn. The structure of deep learning model is highly similar to a human brain with large number of neurons and nodes like neurons in human brain thus resulting in artificial neural network. In applying traditional machine learning algorithms we have to manually select input features from complex data set and then train them which becomes a very tedious job for ML scientist but in neural networks we don’t have to manually select useful input features, there are various layers of neural networks for handling complexity of the data set and algorithm as well. In my recent project on human activity recognition , when we applied traditional machine learning algorithm like K-NN then we have to separately detect human and its activity also had to select impactful input parameters manually which became a very tedious task as data set was way too complex but the complexity dramatically reduced on applying artificial neural network, such is the power of deep learning. Yes it’s correct that deep learning algorithms take lots of time for training sometimes even weeks as well but its execution on new data is so fast that its not even comparable with traditional ML algorithms. Deep learning has enabled Industrial Experts to overcome challenges which were impossible, a decades ago like Speech and Image recognition and Natural Language Processing. Majority of the Industries are currently depending on it , be it Journalism, Entertainment, Online Retail Store, Automobile, Banking and Finance, Healthcare, Manufacturing or even Digital Sector. Video recommendations, Mail Services, Self Driving cars, Intelligent Chat bots, Voice Assistants are just trending achievements of Deep Learning.

Furthermore, Deep learning can most profoundly be considered as future of Artificial Intelligence due to constant rapid increase in amount of data as well as the gradual development in hardware field as well, resulting in better computational power.