Deep Learning – The Rise of AI

Courtesy+of+The+Economic+Times

Courtesy of The Economic Times

Pyae Sone Hmine, Staff Writer

You may have been captivated by the ability of cars to drive autonomously or computers to explicitly hear and see the world around us and even respond to it. But, how exactly is this accomplished?

The key to such mysteries is Machine Learning, computer algorithms that automatically improve themselves through experience.         

Deep Learning is a more complicated Machine Learning which utilizes neural networks, the interconnection of nodes. Deep Learning is a big part of why AI has become popular nowadays, hence the topic of this article. This article will unpack the building blocks of Deep learning, how it works, and its advancements in the world today.

 How would a Deep Learning model learn from experience?

Consider building a learning model that predicts if a given image is a cat or not. First, we will have to tell the computer what a cat is. Since we are teaching the computer by experience, it’s best to show the computer many cats. And that’s basically how we do it.

We give the computer hundreds or thousands of cat pictures as well as other animals so that the computer can later determine if an image is a cat or not. The pictures that we give the computer are called “training data” as it is used to train the computer to be able to perform predictions.

The amount of data available is a great factor in why Deep Learning is becoming increasingly popular. From social media to the web, there is no deficiency of data these days. There are state-of-the-art deep learning models using hundreds of thousands to even millions of data.

Training hundreds of thousands of data takes time. Fortunately, computers are also rapidly increasing in computing power. The computer you currently have is probably far more powerful than a computer 20 years ago. This is also a huge factor in the advancements of Deep learning.

It’s amazing in retrospect to think that computers are just predicting how to drive a car, what a cat is, or what we are saying. But in some sense, we humans are just predicting as well. If you’ve ever seen one of those videos of realistic cakes of everyday objects such as a coffee cup or a book, you’d know that you wouldn’t be able to tell that it’s a cake visually.

Deep Learning’s capability has surpassed just classifying images; however, image classification is still widely used. Cars are now able to maneuver themselves from obstacles and calculate it’s proximity with it’s surrounding cars which are also moving. Elon Musk, the Founder and CEO of Tesla, is positive about Machine learning being used in cars as to using sensors like Lidar systems. Virtual assistants like Google Assistant and Alexa are trained using speech recognition algorithms, which is also basically deep learning.

The uses of deep learning have no limits. It’s a rapidly increasing field. With the help of the advancements of other technologies, there’s no doubt that this technology will expand in the near future.