Owlin x ING x Valohai: Application, Theory & Tooling for ML + Drinks with a View

Owlin welcomes you to a non-standard edition of the Amsterdam Applied Machine Learning Meetup. Together with ING Wholesale Banking Advanced Analytics and Valohai, join us for pizza, diverse machine learning talks and drinks with the best view of the city.

This evening we have three speakers: Floris Hermsen of Owlin, Nanne Aben from ING WBAA and Juha Kiili from Valohai.

Agenda estimate:

17.45 - 18.30 Walk-in & Pizza
18.30 - 18.40 Welcome by Owlin

18.40 - 19.00 Improving Entity Detection & Linking by leveraging Local Context at Owlin

19.10 - 19.30 Combatting Covariate Shift with Adversarial Weighting at ING WBAA

19.40 - 20.00 Asynchronous Model Training & Version Control for Jupyter Notebooks with Valohai

20.00 - ??? Drinks with a view

Floris’ talk description:
Neural networks have become rather good at extracting named entities from texts. But sometimes, rather good is not good enough. How can we leverage statistics, local context and external data sources to boost our detection confidence? We present an existing solution as well as a look-ahead.

Nanne’s talk description:
I will talk about covariate shift: when the data on which we train (for which we have labels) and the data on which we predict (for which we want to infer the labels) come from slightly different distributions. How can we make sure that, despite these differences in distributions, our predictions are still accurate? And importantly, how can we make this scale to problems in which we have ~10M samples and ~400k features.

Juha’s talk description:
Valohai is an end-to-end Machine Learning Platform that takes care of MLOps, automatic version control, and team collaboration. While notebooks are great for early experimentation with ML models, they are not ideal to facilitate company-wide, transparent and production-scale ML development. Jupyhai is Valohai’s Jupyter Notebook extension, letting data scientists run their notebook experiments asynchronously on the cloud while having automatic versioning for code, data, hyperparameters, and more. Time to take your Jupyter game to the next level!