PyConSK 2019, the fourth annual conference for Python developers hosted by PyCon SK, was held last week from March 22–24 in Bratislava, SK.
MindsDB CTO, Jorge Torres, was asked to give a presentation on machine learning democratization at the conference.
We used this opportunity to fly the whole team to Bratislava to attend the conference and hear Jorge’s talk in real-time.
In his speech, Jorge discussed how we’re at a point in time where machine learning needs to be democratized for organizations to be able to make the most of its capabilities. He furthered argued that it’s too much to expect machine learning (ML) experts to have true ML expertise while also maintaining domain expertise across other fields. On this point, Jorge challenged the audience, as well as everyone interested in the field of machine learning, to rethink ways that they and their organizations can build tools that enable even people who aren’t ML experts to utilize the power of machine learning to become participants in the data science community.
Jorge’s talk was divided into three sections:
In the first section, he dove into the importance of democratizing machine learning and how its objective is to demonstrate why it’s more important than ever for the issues around machine learning democratization to be solved. He also provided examples of the risks and problems that appear when machine learning engineers take on the responsibility of building predictive technologies in domains where they don’t necessarily have the greatest expertise.
During the second part of his talk, Jorge focused on providing some solutions and approaches to solving the challenges of democratizing machine learning. As an example of ways to solve some of these challenges, Jorge touched on the journey and results we have seen at MindsDB as we’ve been committed to this endeavor since our inception. He also offered new ways that the ML community can think of and assess the next generation of tools being built.
In the third and most extensive section of his talk, Jorge discussed the new issues that are born when machine learning capabilities are placed in the hands of people who aren’t true ML experts. Additionally, Jorge touched on the implications of delegating machine learning machinery to a system and what has to be true for those systems to be trusted, which led him to discuss the importance of further developing our understanding of explainability, interpretability, and robustness of ML.
Jorge presented to a packed room filled with curious Pythonistas, many of whom made sure to approach him afterwards to discuss the ideas he presented during his talk.
PyConSK 2019 proved to be a phenomenal event for our team to attend because it allowed us to meet and connect with many people from around the world who share our passion for democratizing machine learning.
PyConSK was a very exciting event that gave us the opportunity to not only learn about the ways that other organizations are utilizing Python to build interesting application, but also allowed us to meet new people working on exciting projects and spend quality time with our whole core team.
We also had a ton of fun at a special reception hosted by PyCon SK on Saturday night.
We thank conference organizers PyCon SK for a phenomenally executed event and look forward to attending future PyCon SK events!