PyData Triangle May 2021 Meetup

PyData Triangle welcomes you to another exciting event.

This will be an online event. You must RSVP to this meetup event in order to see the Zoom URL. If prompted, the password is 509458

Speakers:

• Vikas (Vik) Menon
• Nisarg Raval
• YOU: Lightning Talks (Sign-up for a 5 minute lightning talk slot at the meeting by posting in the chat. Or pre-sign-up by posting a comment into this announcement.)

Schedule:
6:00-6:15 announcements
6:15-7:00 Vikas Menon
7:00-8:00 Nisarg Raval
8:00-8:30 Lightning talks

The PyData code of conduct ( http://pydata.org/code-of-conduct.html ) is enforced at this Meetup. Attendees violating these rules may be asked to leave the meetup at the sole discretion of the meetup organizer.

NOTE: This meeting will be recorded.

Please propose a presentation or speaker for a future PyData Triangle meetup. Contact any of the organizers, Aarthi Janakiramen, Dhruv Sakalley, Gene Ferruzza, or Mark Hutchinson through meetup messages.

Follow us on twitter at: https://twitter.com/pydatatriangle

Presenter: Vikas Menon

Title: Interpretable Sequence Learning for Covid-19 Forecasting

Presentation Overview:
Sercan presents a novel approach that integrates machine learning into compartmental disease modeling (e.g., SEIR) to predict the progression of COVID-19. Our model is explainable by design as it explicitly shows how different compartments evolve and it uses interpretable encoders to incorporate covariates and improve performance. Explainability is valuable to ensure that the model's forecasts are credible to epidemiologists and to instill confidence in end-users such as policy makers and healthcare institutions. Our model can be applied at different geographic resolutions, and we demonstrate it for states and counties in the United States. We show that our model provides more accurate forecasts compared to the alternatives, and that it provides qualitatively meaningful explanatory insights.

Bio:
Vik has worked as a software engineer for 12 years in various capacities. A common theme across all his roles has been large data.

In his current role he solves Google's cloud customers AI problems particularly focussed in the public sector space. His last project involved building a novel ML augmented compartmental model for predicting the spread of Covid-19 in the US and Japan.

He spends most of his time hunting bugs in code and inconsistencies in data. In his spare time you will likely find him obsessing over weightlifting or riding a bike while listening to podcasts

Presenter: Nisarg Raval

Title: Protecting Secrets using Adversarial Nets

Presentation Overview:
Personal data collected from various smart devices are often offloaded to the cloud for analytics. This leads to a potential risk of disclosing private user information. On the other hand preventing apps from using user data hinders their functionality. In this talk, I present Olympus, a privacy framework that helps users balance privacy-utility trade-offs, and allow them to make an informed decision about sharing their data to apps in order to obtain services in return. Olympus achieves privacy by designing a utility aware obfuscation mechanism, where privacy and utility requirements are modeled as adversarial networks. Through experimentation on a real world app and on benchmark datasets, we show that Olympus successfully limits the disclosure of private information without significantly affecting functionality of the application.

Bio:
Nisarg earned his PhD in computer science at Duke University in 2019. After graduation, I joined LinkedIn as a machine learning engineer. Currently, he is a senior software engineer at Emerald Innovations.