fbpx

10 Deep Learning Applications for Investors to Watch

Before tucking into some really cool deep learning applications, we need a bit of context first. Probably the most intriguing and exciting technology today is artificial intelligence (AI), a broad term that covers a swath of technologies like machine learning and deep learning. As investors, our ears perked up when we first heard about AI and we immediately wanted to get a piece of that action.

So we did a bit of research and as it turns out, you can’t invest in AI technology yet as a retail investor because there are no pure-play AI stocks yet. Of course this is not what mainstream media publications would have you believe. Editors tell their writers to write about AI stocks and they have to come up with something, so they just start using name associations to come up with some “AI stocks”. We always see these “how to invest in AI” articles demonstrate the “invest in everything with Google” fallacy that assumes you can invest in every single technology known to man by just buying some Google shares. The truth is, even though the likes of Apple, Facebook, and Google are snapping up AI startups left and right, they are still just companies with a core focus on something completely unrelated to AI. There are no pure-play artificial intelligence stocks out there for retail investors yet, though there is tremendous interest in these areas as seen below:

interest-in-artificial-intelligence-machine-learning-deep-learning
Source: Google

So what can retail investors do? The best plan is to become acquainted with all the startups in the artificial intelligence space. We want to get an idea of who is doing what, and which startups are potential contenders for a future IPO. We also want to get very specific about which area of artificial intelligence we’re interested in. The below diagram best shows how all the pieces fit together:

deep learning applications
Source: Nvidia

As you can see in the above diagram, we’re interested in learning more about deep learning because that’s what will ultimately power all the real breakthroughs that we’re going to see in a very short period of time if we’re to believe the hype. In order to eventually invest in deep learning, we first need to understand all the deep learning applications that are available. Let’s break down 10 of the most promising deep learning applications found across various industries and provide some specific examples of startups actively playing in these spaces.

Deep Learning in Robotics

Probably the most exciting deep learning application is in the area of making smart robots. We all have visions in our heads of humanoid robots walking around and doing useful things, like our chores for example. Since they stole all our jobs it’s the least they can do to pay us back. If you take this down a few levels though and think about it, autonomous cars are robots of sorts that use deep learning to navigate and most importantly, they use deep learning to actually learn. Check out these 5 new self driving car companies that are all using deep learning to help pilot autonomous cars. Just to show you how new this application of deep learning is, none of these companies are more than 2 years old.

Deep Learning in Affective Computing

What good is it to have a robot walking around acting like a humanoid if it can’t tell when you’re hungover and not in the mood for small talk? The use of deep learning to sense human emotion is called affective computing and it’s something marketing people are drooling all over each other about. Detecting human emotions is particularly applicable to seeing how people react to visual stimuli such as ads or commercials. These days everyone gets offended over everything so you can use affective computing to make sure that doesn’t happen. One company playing in this space is Emotient, and their business model is to charge advertisers to analyze how consumers respond to ads. Eventually we’d expect to see voice applications as well.

Deep Learning in Healthcare

This domain will perhaps be one of the most fruitful for humanity. If a deep learning algorithm can beat a human at a game of Go by spotting patterns, then pattern spotting might just be where we see deep learning used first in the area of healthcare. We talked before about a company called Enlitic that is using deep learning to read X-rays better than a radiologist and this technology is available now. Even more exciting than that is the application of deep learning for drug discovery, and there are at least 4 startups working on this right now.

Deep Learning in Cybersecurity

We wrote before about the topic of cybersecurity and how there are major investments being made in this space because having a secure environment is not an option. It’s like your health. Companies will invest in whatever it takes to make sure they’re not the next poor sap to end up on the nightly news having to admit that someone stole all their client information. One application for deep learning in cybersecurity is pattern recognition of viruses or what they call “virus signatures”. The reason your anti-virus software is always updating itself is because it needs to go get the latest “signatures” that it can use to recognize new viruses. One startup called Cylance is developing deep learning algorithms that can live on your laptop and with no internet connectivity, dynamically detect virus signatures for new viruses. Now that’s pretty cool.

Deep Learning in Genomics

Examining the digital genome is a very data intensive activity. If you just took the letters of one person’s genome and wrote them down, that data would take up an astounding 700 megabytes. If you took the real-world genome data right off of an Illumina genome sequencing machine, that data file would be around 200 gigabytes. Isn’t that incredible? We’ve learned that in order to train deep learning algorithms, you need to feed them lots of delicious big data. Genomics turns out to be the perfect domain for deep learning applications and one company Deep Genomics is doing just that by examining the effects of gene editing.

Deep Learning in Finance

We have to admit that as investors, the first deep learning application that crossed our mind was stock trading. If these neural networks are so darn good at mining massive big data sets, then why not turn them loose on the stock market and have them trade our portfolios? As it turns out, that ship sailed a long time ago and some of the world’s most successful hedge funds are already using deep learning to generate some delicious alpha. Another interesting company in this space is Signalfire which is using unstructured data from 2 million data sources to pick which startups they invest in. Check out our article on 8 hot startups they picked.

Deep Learning in Computer Vision

In a previous article we gave you an example of how you can train deep learning algorithms to recognize what is in a photograph. Essentially you just feed the algorithm tons and tons of pictures and train it how to “see”. The algorithm will then present pictures that it’s having a hard time recognizing and a human can then help clarify any ambiguities. Here are 5 computer vision and image understanding companies that are making progress in this space. Eventually, deep learning algorithms will be good enough to just recognize everything they “see” in real-time with almost no lag.

Deep Learning in Conversational Interfaces

Some day in the future you should be able to talk to a computer in the same way as you talk to a human without having to adjust your speech in any way shape or form, and the computer should be able to understand you at the same level as a human. We took a look at some of the best voice recognition technology out there and were not really impressed. As far as chat bots go, that experience was even worse. Conversational interfaces are a domain that is ripe for deep learning to address, and we recently profiled a company called Viv that is working on this very same problem. They were doing such a good job that Samsung recently acquired them.

Deep Learning as a Platform

Deep learning itself can be industry agnostic such that a company can market their offering as a platform as opposed to targeting any one particular industry. We see a good example of this in a startup called “Sentient Technologies” which has developed a deep learning platform which solves problems using a form of natural selection called “evolutionary intelligence”. Sentient is targeting three different industry applications so far; finance, healthcare, and retail.

Deep Learning and Quantum Computing

Ok, so we admit we’re not actually aware of any companies using deep learning on quantum computers yet but we wanted to toss this one out there because this is the coolest deep learning application we could dream up. Loyal readers will recall an article we wrote on 5 companies building artificial intelligence (AI) chips  where we talked about how a special type of hardware is needed for AI. This is why companies like NVIDIA who specialize in graphics chips are suddenly being looked at as “picks and shovels” plays on AI. Heavy lifting processors like the latest from Intel just don’t play well with AI. So what the heck happens when you run a deep learning algorithm on a quantum computer? Maybe one of our lovely readers can chime in here with some thoughts on what future deep learning has with quantum computing (if any).

Conclusion

So we’re all done now with our primer on 10 deep learning applications that investors should be watching. For retail investors, all we’ve told you is that you can’t invest in deep learning yet because there are no pure-play deep learning stocks. That may not seem very helpful, but if you ever plan to invest your hard earned dollars in a deep learning IPO, then you’ll be much more informed after learning all about the 10 deep learning applications we have described.

Share

Leave a Reply

Your email address will not be published.

  1. Nice post! I am organizing a deep learning hackathon at the Kiev Polytechnic Institute, best scientific university of Ukraine. The topic is deep learning applied to real-world problems, and your post is really inspiring for our preparation!

    I am looking for remote speakers to present their challenges to our talented crowd.

    If your fund/company has cash to spend on real-world problems in AI and deep learning, anywhere in the world, contact us!
    We hope this event will initiate new collaborations: recruitment, funding, consulting, outsourcing…

    1. Thank you for that Mostapha!

      Coincidentally, the author of that article was in Uzhhorod last year on holiday and is looking to buy property in Lviv. We told him to look you up when he’s in the region!