Machine Learning Algorithms are Tools
There are some really amazing advances and some phenomenal machine learning algorithms and deep learning algorithms, but those are just tools that need to be applied. Companies take those core machine learning tools–they don’t have to build everything from scratch–then they apply the machine learning algorithms to a specific problem. So in our case, understanding the nuances in subtext and sarcasm must be done with a human-in-the-loop. There is no machine learning tool that does that for data scientists, but there are tools that allow us scientists to mine data more effectively.
Our core IP is the secret sauce. We outsource development a small amount of solely the basic reporting functions. But the core IP that the models are running to generate results are designed in house. Our team of data scientists is busy investigating and tweaking those algorithms. The learning curve is just too steep to outsource development of our core intellectual property to anyone.
It’s difficult not to build some customization into the core infrastructure and get a little bit stuck in the details when our clients talk about analytics as a service. We built our platform to be plug and play way so that we can go through a dozen of APIs that are doing sentiment analysis, entity extraction, spelling grammar punctuation test. I can tell you about who is doing the best personality analysis although I won’t do that publicly. We’ve tried IBM Watson, Google, and Microsoft. We knew we had to mix and match and do this the right way so we built this from the ground up knowing that that was really important.
Dr. Michael Housman, Chief Data Scientist and Co-Founder of RapportBoost.AI, discusses using new and existing tools and API to be more effective in creating better solutions at the JMP Securities Conference panel on AI.
Learn more about the live chat sales solutions from the awesome team at RapportBoost.