The KDD 2019: The annual rendezvous for data science experts

By Hugo Roucau-Montagné

Peter Lee, Corporate Vice-President at Microsoft Healthcare, was one of the notable KDD 2019 speakers.

The Knowledge Discovery and Data Mining conference is an annual event where data science experts from all parts of the world gather together for five days to share their best practices and latest advances. This year, more than 2,500 confirmed attendees traveled to Anchorage, Alaska to participate in highly collaborative lectures, workshops and hands-on tutorials. Among others, I had the chance to listen to Cynthia Rudin from Duke University, Elena Grewal, Head of Data Science at Airbnb, and Peter Lee, Corporate Vice-President at Microsoft Healthcare.

Even though the spectrum of topics addressed was extremely large, there were two concentrations for the KDD 2019: Deep Learning and Data Science for Environmental Impact.

Overall, it was an amazing, enriching experience, as industry experts as well as academics shared in-depth insights about their research. Most of the attendees brought a piece of work to present which created a very open and sharing atmosphere during the event.

My classmate, Pierre Melikov (MS student in Civil Engineering / Systems), and I had the pleasure to present a project conducted while I was in the Berkeley MEng program.

My classmate, Pierre Melikov (UC Berkeley MS student in Civil Engineering / Systems), presented a project we conducted in partnership with the AI for Good Foundation. Our project addresses the issue of bias in written news content. We focused on quantifying the bias of a newspaper from a holistic perspective based on the analysis of its articles to enable news consumers to quickly form an opinion regarding the objectivity of a news outlet. A machine learning model was trained to predict the bias of a written piece, regardless of its source. Then, the results were aggregated at a newspaper level and were confronted with a survey conducted by Gallup and Knight Foundation. The model successfully captured what the surveyed population identified as ‘bias.’ Further work is still required to launch it and enable public access, but the initial results are very promising.

It was an extremely valuable experience to be challenged by some of today’s most brilliant people in Data Science. The KDD 2019 conference was a unique experience I warmly recommend to all data science enthusiasts. Luckily, KDD 2020 will take place in San Diego!

--

--

The Fung Institute for Engineering Leadership
The Fung Institute for Engineering Leadership

Written by The Fung Institute for Engineering Leadership

The Fung Institute for Engineering Leadership at the UC Berkeley College of Engineering is reinventing engineering education for the digital age.

No responses yet