What is Deep Learning And How Is It Affecting The Future Of AI?

The idea of artificial intelligence that can think just like a human is as old as the idea of artificial intelligence itself.

Deep or Neural learning is a specific subset of machine learning which is focused on the process by which artificial intelligence learns to think like the human brain by using a neural network inspired by it, known as an artificial neural network or ANN

Unlike traditional machine learning, deep learning doesn’t rely on the typical commands necessary for traditional machine learning and instead is able to “think” about the data on it’s own and extrapolate insights that humans might not even thought to have asked for.

To use a sports example, machine learning can be used to figure out pitch framing or the impact that a catcher has on the calling of balls and strikes for a given pitcher. Machine learning can given us the probability that a will be called a strike. Deep learning can then tell you which catchers are the best or worst at framing pitches and which of the pitcher’s pitches will be mostly likely to get called balls or strikes by extrapolating insights from that data allowing the pitcher to select the pitches most likely to be called strikes before a game based on his catcher.

Why is Deep/Neural Learning Important?

Deep/Neural learning represents the next step in artificial intelligence evolution.

In order for companies to achieve true Cognitive Automation where machines are able to analyze data and then extrapolate insights without human analysis, they must develop deep learning to the point that the machines can think like humans.

What are some industries that are currently using deep/neural learning?

  1. The Legal Field. Casetext is using deep learning to help lawyers check their legal briefs and those of their oppositional council. The user simply uploads his or her brief as well as the brief of their opposing council and Casetext instantly checks it for key citings, missed precedents and more.
  2. The Medical Field. Recursion Pharma uses bio rich data extracted from medical images to search for new drugs that may treat rare disorders. For most rare diseases the timeline for drug discovery is around ten years at a cost of around 1 billion dollars. With recursion technology they have mapped almost 100 diseases in the last quarter alone with the goal of being able to shorten the drug development pipeline with deep learning.
  3. Voice Interactivity. At the CES conference in January one of the big stars was Houndify voice activation platform which was featured in everything from a voice activated personal assistant inside of a Hyundai to a voice controlled robot which danced like Michael Jackson. Houndify open source platform with voice development allows for even more developments which will use deep learning to continue to develop voice interactivity that talks and thinks just like a person.
  4. Workflows. One key area where deep learning is already helping businesses run more efficiently is in digitizing operations and eliminating manual work. Companies like WorkFusion are providing level automation, digitization and cognitive AI to help processes and workflows run smoother and more efficiently.

What are the hopes for the future of DL?

While it’s still early, some of the companies I spoke to are hopeful that we will see a few of the following breakthroughs over the next few years.

  • `A complete biological mapping of both the human body and the compounds found in common therapeutic drugs. This would allow for a two pronged approach to precision medicine and precision therapeutics using deep learning.
  • Voice activation in the Internet of Things (IOT). Open source platforms like Houndify and “Smartbrains” created by companies like Brain Corp are going to allow for the adding of features like autonomous movement and voice activation to a variety of real world items like coffee machines.
  • Fully automated trials and sentencing. While right now AI is mostly being used for research and safeguarding from legal consequences there has been a good amount of research being done at institutions such as Stanford that suggest that deep learning could be used to teach AI to interpret and recommend sentencing in court cases.

Deep Learning is the process of creating an artificial neural network (ANN) which will structure data and algorithms similarly to the human brain.

Deep learning is important because it represents the future of AI and ML

It’s already being pioneered by companies like WorkFusion and Brain Corp to start harnessing it’s incredible power in industries like law and autonomous machinery.

In the future it is believed that Deep learning is the key to developing things like

  • A complete map of human biology and drug compounds allowing rare diseases to be treated faster and cheaper throughout the drug development pipeline.
  • Legal decisions and sentencing without bias or human judges.
  • Improved voice recognition technology which can predict what you want from context clues.

The future of AI is being developed today!

Deep learning is the key to developing the next generation of AI that understand context and language in a way that will unlock the future we have always dreamed of.