Deep learning is entering the business world, and the economy is beginning to feel the impact. Worth $272 million in 2016, the deep learning market is growing at an annual compound rate of 52.1 percent, putting it on track to reach a value of $10.2 billion by 2025, Grand View Research estimates.
As more companies deploy this revolutionary technology, deep learning will have an increasing impact on the economy as a whole. Here’s a look at what deep learning is, how innovations in the field are making it more practical for businesses and how these developments are setting the stage for profound economic transformation.
Defining Deep Learning
Deep learning is one of today’s most successful applications of machine learning, which is in turn a cutting-edge application of artificial intelligence. AI uses computers to replicate human cognitive tasks such as logical deduction, mathematical computation and language processing. One goal of artificial intelligence research is develop an artificial brain that can replicate any human cognitive task, a vision known as generalized AI. However, most AI technology focuses on performing specific tasks.
Machine learning is a specialized AI application that uses probabilities to analyze data in order to identify mathematical patterns, make predictions or determine the most efficient method to perform a task. Due to this reliance on probability, in contrast to traditional AI programs, machine learning is not necessarily locked into a preprogrammed routine, but can explore a range of ways to apply algorithms to data.
Deep learning integrates machine learning with artificial neural networks modeled on the human nervous system. Just as the human nervous system receives and processes sensory input, neural networks receive digital input, assign logical or mathematical values to the input that represent how closely it matches to a desired outcome, classify the results and use the results to generate output. This process can run through multiple iterations until it reaches an outcome that approximates a desired goal. The results can be used to identify trends, build predictive models or make decisions.
For example, one application of deep learning is computer vision. A deep learning application might be assigned the task of seeing the image of a dog by comparing known images of dogs with digital input from artificial sensors to determine how closely the sensed input matches the visual pattern of a dog. Deep learning can be used in a similar way to find patterns in images, language or consumer buying trends.
Making Deep Learning Practical
Neural networks have been around for decades, but until recently, they drifted off from the main body of AI research due to the immense computational power required, which made them impractical. However, recent advances in data storage and processing capability have finally begun to make deep learning practical over the past five years.
The cloud has helped open up machine learning and deep learning to businesses by making remote computing resources available on a scale far larger than previously possible. Meanwhile, as smartphones have increased in processing capability in order to handle high-speed internet connections and large amounts of photo and video data, mobile devices have developed the capability to run on-device AI, machine learning and deep learning applications without relying on the cloud. For instance, Qualcomm’s new Snapdragon Artificial Intelligence platform can use deep learning to optimize photos, eliminate background noise from audio or recognize user faces for biometric authentication, among other applications.
Applying Deep Learning to Economics and Business
On a macroeconomic level, one of the most far-reaching effects deep learning will have on the economy is empowering more accurate economic models. An international team led by German researcher Christopher Krauss has already demonstrated that artificial intelligence can be used to make profitable stock market predictions, with AI S&P 500 selections from 1992 to 2015 generating double-digit annual returns.
On a microeconomic level, deep learning can also be used to generate more accurate models to identify marketing and sales trends and improve operational performance. Amazon and Microsoft recently teamed up to release an open source deep learning platform called Gluon that makes it easier for companies to develop their own AI applications.
One of the most profitable applications of deep learning for businesses is marketing and sales. For instance, Salesforce’s Einstein platform applies deep learning to CRM data to allow companies to predict customer buying trends, identify hot prospects and suggest products that are likely to appeal to individual buyers.
As more companies adopt these types of deep learning applications, this revolutionary technology will have an increasing impact on the fortune of individual businesses and the economy as a whole, with both microeconomic and macroeconomic decision-making becoming more data-driven.