For driverless cars to work safely, they need to be able to do things like recognize signs, distinguish pedestrians from other objects along a roadway, and understand traffic patterns. There has been a lot of excitement over the possibility that deep learning can give driverless cars those capabilities. But if your ISV isn’t developing solutions for autonomous vehicles, does deep learning have potential benefits for you and your clients?
As with most emerging technologies, hype has built up around deep learning, which, according to the Gartner hype cycle, sets it up for a fall into “disillusionment” before it becomes mainstream. Forward-thinking ISVs don’t have to take that rollercoaster ride with popular opinion, however. It may be the perfect time to consider what this technology can do for your clients, to learn and develop your skills, and to build a strategy to include deep learning in your solutions.
What’s the Difference Between Machine Learning and Deep Learning?
You may have heard the two terms used interchangeably, but machine learning and deep learning differ. With machine learning, algorithms learn from data that’s given to them. Examples of machine learning include a streaming service that makes music or movie recommendations based on other user’s behaviors or identifying an object from a photo by matching specific characteristics.
Deep learning is a type of machine learning that gives computers the ability to attain higher recognition accuracy – in some cases exceeding human performance – and as data volumes increase, it can actually improve its performance. Deep learning uses neural network architectures that can be trained to learn features directly from an image, speech, video, or other data, so you don’t manually have to extract the features for it.
Deep Learning Applications
So it’s easy to see how deep learning could advance the way driverless cars work. It could enable them to extract information directly from the things, signals, and movement around them, and to continually learn by interacting with more data. But it’s value doesn’t stop with driverless cars. It can also make a big impact in other applications:
- Healthcare
Deep learning can aid researchers to discover new medicines and accurately read scans to diagnose cancer and other diseases or conditions.
- Retail
Voice-activated assistants can recognize the user’s voice, learn their preferences, and help them shop.
- Translation
Deep learning can enable automatic translation of text, whether written or spoken, to another language.
- Defense
Using data from satellites, deep learning can identify objects or activity and detect areas that are unsafe.
- Safety
With deep learning, a system can automatically detect unsafe conditions and alert workers or shut down equipment. It’s also proven to be much quicker and more effective at predicting earthquakes.
- Cybersecurity
Deep learning can shorten the time it takes to detect new malware activity or intrusions. It can also interpret email text to more accurately classify it as a phishing attempt.
- Identification
Systems could have the ability to identify people, interpret their expressions and recognize gestures. This could be the key to allowing people to automatically pay for goods and services or gain access without having to show a card or other credentials. And if you need to zoom in for positive identification, deep learning can help overcome low resolution with pixel restoration.
- Robotics
Deep learning can provide robots with vision and touch to make them capable of more tasks, both in an industrial setting and a household.
Deep Learning for Beginners
There are a number of on-ramps into deep learning, including internet resources, like Analytics Vidhya’s post on deep learning applications you can build using Python. This article includes examples like image tagging, apparel recommendation, and filtering inappropriate images.
You can also leverage preconfigured environments to build deep learning applications, such as Deep Learning AMIs offered by Amazon Web Services.
If you can see beyond the hype to the value deep learning can provide, you can begin now to devise a plan for adding new capabilities to the solutions you develop – which could be real game-changers for your clients and your business. 
