8 Lessons From Big Data Toronto 2017 – AI Is Everywhere

Last week I attended the Big Data Toronto 2017 – 2 days conference and tradeshow at Metro Toronto Convention Center. My impressions from this year’s conference were significantly different from the feelings I had after attending the first Big Data Toronto in 2016.
I wrote about my impressions from the first conference in my blog article – Big Data Is Still Like Teenage Sex – Big Data Toronto 2016

Many presentations of last year’s conference were about why do we need Big Data and how to start exploiting Big Data. Some of the speakers presented more traditional business intelligence and analytics topics. Only a few stories were touching real Big Data business applications.

I was pleasantly surprised that many presenters at Big Data 2017 were showcasing examples of leading edge Big Data solutions already working to benefit their business.
Some of the speakers represented the new generation of companies like Uber that have Artificial Intelligence technologies or The Internet Of Things (IOT) at the core of their operations. There were also examples of more traditional organizations working in financial, retail, healthcare industries or government.

1. More data is making us dumber

One of the keynote speakers quoted this phrase. The reason is that our ability to understand data is developing much slower than the volume of data. We need help from AI technologies, particularly machine learning to bring the situation under control.

2. Data and algorithms become significant competitive advantage

Ability to understand data faster and better than competitors is the primary advantage of the new post – industrial economy. One of the speakers even called that Amazon is an algorithm company rather than the biggest online retailer. Amazon was able to apply smart algorithms in almost all areas of their front-end and back-end operations.

3. Traditional boundary between industries are rapidly changing

The new generation of innovative companies like Facebook, Google, Amazon, Apple or Uber were able to use their advantage in big data expertise to become major players in different industries such as finance, retail, home automation, or transportation. Mastering big data is a necessary skill for any business to continually reinvent itself and survive in the data-driven economy.

4. Artificial Intelligence (AI) technologies take center stage

I heard this phrase multiple times during last 12 month. This fact was also very noticeable at Big Data Toronto 2017 conference. This year many presenters talked about solutions that utilize AI technologies. On contrast, during 2016 event only a few presenters mentioned AI related technologies. The MapReduce model and advanced visualization techniques grabbed most of the audience attention at last year show.
I regularly follow the new development in Artificial Intelligence technologies and fully agree with the statement “A.I. is advancing faster than many people may think”.

5. Machine learning is everywhere

Companies that successful master Big Data can apply Machine Learning technologies to a wide range if business use cases:

• Churn Prediction
• Customer clustering and segmentation
• Product recommendations
• Forecasting ETA (estimated time of arrival)
• Sales predictions
• Price optimization
• Fraud prevention
• Risk forecast and management in many industries
• Surveillance, physical and cyber security
• Healthcare: disease propensity, modeling ICI occupancy, hospital readmission risk
• Pharmaceutical: drug delivery optimization.

6. Deep Learning is hot

It was entirely predictable that deep learning was a trending topic during the conference. It became a compelling technology in the last ten years due to the combination of multiple factors; such as reduced cost of computer power and storage to enable the learning, advancement in algorithms, and the enhancements in our ability to produce digital data from different sources or sensors. The advantage of deep learning that it can be applied successfully to a vast range of pattern recognition problems. One of the speakers has shown an example how the computer was learning from Ernest Hemingway novel to write meaningful English text and with the writing style that is closely similar to the original.

While deep learning is making fast progress in recent years, it is not a silver bullet to solve all machine learning problems. When we select a potential solution for the real life problem, we have to consider the economic factors. Deep learning requires specialized processing hardware (GPU) to build production models from large volumes of data, and that makes the project cost quite expensive.

7. Machine learning still needs people (data scientists) to select the right tool

We should match our tools with a particular challenge rather than consider deep learning as the only universal solution for any problem. This message came across in few presentations during the conference. It may be cheaper and faster to apply simple linear regression or SVM to solve the particular problem.

8. The fastest progress is happening at the intersection of multiple technologies

It was visible that the most rapid growth is occurring at the intersection of multiple technologies such as the Internet of things (IOT), big data and AI. We had the opportunity to learn about use cases in fleet management, healthcare, retail, environmental protection, smart homes and smart cities.