Proceed with caution before diving into deep learning investments. It's critical to do your homework and only allot 5 to 10 percent of your big data budget to deep learning, if you decide to try it.
Google's recent $500 million acquisition of DeepMind, and Facebook's recent hiring of NYU professor Yann LeCun, a respected pioneer in artificial intelligence (AI) and deep learning, has a lot of people talking about deep learning. So is deep learning a good fit for your corporate strategy? In short, it depends on what you're trying to accomplish.
Deep learning: What every leader needs to know
Think of deep learning as cutting-edge AI that generally represents an evolution over primitive neural networks. A key distinction between traditional machine learning and deep learning is the amount of supervision and human intervention the AI system requires.
Traditional machine learning techniques, including classic neural networks, need to be supervised by humans so they can learn. Deep learning is an attempt to have the system learn on its own, without human intervention. It may sound like science fiction and rather far-fetched, but we've actually experienced success in certain areas using deep learning, such as image and voice recognition.
Maybe the most compelling reason to consider deep learning for your organization is if it supports a breakthrough strategy, where you can be competitive or distinctive with traditional machine learning.
For instance, imagine that you're trying to decode the digital behavior of customers. You're in a very competitive space, and your competition is already making progress in the same pursuit with its big data analytics. You don't know what techniques your competitor is using, but you do know their team of data scientists is busy working on advanced algorithms, so it's a safe bet that machine learning is part of their arsenal. A competitive strategy is not an option here—you need to build something fantastic that your competition can't match. This is a good time to roll the dice on deep learning.
Deep learning shouldn't take deep pockets
I believe deep learning is a good investment; however, you need to do some homework first. Because of the all the noise about deep learning, this homework won't be as easy as you might think; you'll either need to get outside advice from someone you trust or carefully process all information through a business leader's filter.
Whatever you do, don't turn the decision over to your data scientists. This is far too intellectually attractive for a typical data scientist to give you an unbiased opinion. You need to make an economical decision—you're running a company, not a research facility. The problem, though, is the subject matter gets thick pretty fast, so it really does take a data scientist to fully understand the science and technology that's under consideration. So, unless you know someone who has a background in data science and the mindset of a business person, you'll have to rely on your business sense to make sure you're not getting into a bad situation.
If you've done your homework at a high level, you should already know your cost threshold for making a smart investment on big data analytics—this will prevent you from making an egregious mistake with your money. However, it's not smart to dump this entire allotment on deep learning. You should think of deep learning as a high-risk/high-reward play.
Take five or ten percent of your budget for big data and invest in a deep learning specialist and maybe a small, dedicated server farm. If your budget can't afford this type of investment, reconsider how you're going to use deep learning in your corporate strategy. You don't have to entirely abandon the idea—just don't make it an integral part of your strategy. Maybe use it as a development opportunity for your data scientists to boost morale and attract top talent. They'll love it.
Deep learning is the catnip of data science these days, but euphoria and hallucinations don't serve as great innovation machines. It's your responsibility to make sure that you make smart investments with your big data dollars. Deep learning is one of those areas that can become a huge money pit with very little return. I'm not suggesting that you stay away from it—just proceed with caution.
The first step is to understand what you're dealing with, so take some time today to investigate. Approach it with an open mind and a business sense. If you find something interesting and you have the budget, take a chance.
Once in a while when you go deep, you score a touchdown.