Develop This: Deep Learning is the Next Seismic Shift
Deep learning is riding high on the next wave of technologies disrupting industries from banking and finance, to healthcare and customer service. To find out more about how developers are beginning to use deep learning, I reached out to a few experts who are among the instructors of a new openSAP MOOC entitled, “Enterprise Deep Learning with TensorFlow.”
Machine intelligence with a human twist
A subset of artificial intelligence and advanced machine learning, deep learning is being partly fuelled by the big-data explosion. Deep learning capabilities allow people to slice and dice the mountains of structured and unstructured data they’re collecting for exceptional results that impact the business. Similar to how the human brain thinks, deep learning can power autonomous cars that “know” a pedestrian has stepped into a crosswalk, banking systems that readily detect money laundering, or help physicians devise more effective treatment plans. Deep learning can also be incredibly effective at making customers smile.
“At SAP, we’ve embedded deep learning capabilities into our IT support process to classify tickets and suggest solutions based on the text in the tickets,” said Karthik Muthuswamy, Machine Learning Researcher at SAP. “This has reduced turnaround time and resolved customer problems faster.”
What’s different now
Huge advances in natural language, speech recognition, object detection and image recognition are solving problems once thought impossible through deep learning.
“Deep learning is the engine of more than AI. It represents this biologically-motivated approach of simulating the human brain on a machine,” said Dr. Damian Borth, Director of the Deep Learning Competence Center at DFKI.
Borth predicted deep learning will have a tremendous impact on many industries, especially automotive, retail and finance. “Autonomous vehicles will change how societies think about mobility in the future,” he said. “Last year, we had a run on the retail industry when the Amazon Echo device recognized the consumer’s voice which was translated to orders. This year the financial industry is looking for advice on deep learning, and my intuition tells me next year will bring MIFID II regulatory changes, sparking transparency updates that will bring more innovations.”
A boon to healthcare
Josh Gordon, Developer Advocate on the TensorFlow team at Google, said his favorite deep learning applications were in healthcare, specifically medical imaging. “There are three applications I can think of, all of which are designed to assist physicians, and I’m sure there are many more. The first used TensorFlow to detect eye disease, the second to detect cancerous cells in tissue samples, and the third was a mobile app to detect skin cancer.”
Gordon is particularly jazzed by the emergence of deep learning as a productivity booster. “One of the most remarkable things about deep learning is how accessible it has become to developers,” said Gordon. “In the past few years, we’ve seen this great combination of researchers and developers sharing their ideas and tools, leading to increased productivity for everyone.”
Building deep learning skills
The openSAP MOOC on deep learning that all three gentleman are instructing, is designed as how-to introductory course for developers and data scientists, which will explain when and when not to use deep learning to solve organizational problems, as well as how to build and deploy deep learning models in enterprise systems. The curriculum is heavy on examples of industry applications and hands-on coding lessons, along with an online forum to exchange ideas with other developers worldwide. Participants will receive a certificate of completion after passing a final exam.
Follow me @smgaler