FOLLOW

FOLLOW

SHARE

How Is Deep Learning Revolutionizing Artificial Intelligence

Deep Learning is making a bigger difference to our lives than we may realize

3May

While we often think of artificial intelligence in terms of androids that walk and talk like humans, AI absolutely exists in our modern society. From chat applications that respond to human communication to robots that can complete tasks and alert technicians to errors, our society is already being transformed by AI, to the level that Andrew Ng at Baidu Research referred to AI as the new electricity. Deep learning, meanwhile, is rapidly revolutionizing artificial intelligence.

To understand how deep learning is making such a difference, it is important to have an accurate understanding of what deep learning really is.

What is Deep Learning

Deep learning is a subset of a subset of the way that machines think. Artificial intelligence is the broadest term we use for technology that allows machines to mimic human behavior. AI can include logic, if-then rules, and more.

One subset of AI technology is machine learning. Machine learning allows computers to use statistics and use the experience to improve at tasks.

A subset of machine learning is deep learning, which uses algorithms to train computers to perform tasks by exposing neural nets to huge amounts of data, allowing them to predict responses without needing to actually complete every task.

What Areas of AI Are Being Affected By Deep Learning?

There are a few obvious areas where casual users are seeing a big difference in their technology experience due to deep learning. These include:

  • Speech Recognition. Whether you use Cortana, Siri, or Ok Google, your smartphone or computer’s ability to recognize your voice and answer your questions is a function of deep learning. The software has been trained to recognize the myriad ways a word can be pronounced, then understand the various types of questions that might be asked, and to provide an appropriate answer to meet your needs.
  • Smart Homes. Using data about when your home needs light, temperature adjustments, and more, smart home technology allows customers to reduce energy usage and save money. When combined with voice recognition tools, smart homes can have profound applications for disabled consumers.
  • Image Recognition. If you take pictures on your smartphone and store them in various cloud applications, you may have already encountered apps that will sort and organize albums based on various features. All the images of a particular person, for example, or images that contain a particular theme. But image recognition has dramatic implications in the field of health and wellness. According to some medical startups, computers will soon be able to read your CT, MRI, or X-rays images, and diagnose you without needing a radiologist to interpret the results. The software may be able to diagnose cancer and other medical concerns more quickly, accurately, and less invasively than modern methods.

What Is The Next Step For Deep Learning?

There are two factors right now that companies need to be aware of before they can take full advantage of deep learning.

The first is that the hardware needs for deep learning systems are incredible. Deep learning is performed by neural networks, which are massive computing systems. These systems are then fed terabytes of data in order to teach them what they need to know to perform their functions. This is actually changing the kinds of chips used in systems designed to facilitate deep learning.

Instead of traditional CPUs or central processing units, software developers found that GPUs or graphic processing units, were much more efficient. These GPUs were actually developed to create 3D gaming experiences.

Google revealed that they have been using their own custom-designed TPUs or tensor processing units, to implement its own deep learning applications. Intel, meanwhile, has been purchasing startups that are developing deep learning computation technology.

Deep learning is also an incredibly specialized field, as a subset of a subset within information technology. The average software programmer is unlikely to be completely familiar with what deep learning is, or how to take full advantage of the technology. Companies looking to make strides with their own AI options will need to look for very specialized programmers who have experience in this field.

Over time, one of the primary goals of AI developers is to create software which can teach itself. Right now, programmers still need to show neural nets, for example, that this image is of a cat while this other image is not. In time, it is believed that we will create software that can identify the similarities and differences on its own.

In the meantime, another important field within deep learning is managing human bias. For example, we know that people tend to have different impressions of human faces depending on the color of the facial skin. Since humans are programming deep learning machines, it is possible for those biases to be programmed into the software, which could affect generations. 

Comments

comments powered byDisqus
Data culture small

Read next:

Building A Culture Of Data

i