LDN *Virtual* Talks November 2020 (Data Science Special) with The ODSC

Rust London is happy to announce the final LDN Talks of 2020. We will be having another special edition meetup that focuses on a specific domain of the Rust programming ecosystem. Rust London has teamed up with the ODSC https://odsc.com/ to showcase all the exciting things being done with Rust in Data Science. We will have 3 talks from members of the Rust ML Working Group that will present projects and share insights on how Rust is being used in Data Science.

This event will be Live-Streamed due to the Corona-Virus Pandemic

Agenda:

• Welcome

• News and Announcements

• Speaker: Andraz Tori

• Speakers: Paul Korebitz & Lorenz Schmidt

• Speaker: Daniel Mckenna

-----
Andraz Tori
Is the Head of Recommendations and Data Science at Outbrain, Zemanta co-founder, tech enthusiast and also ex-TV host.

Speed is The Feature: Fwumious Wabbit:

Fwumious Wabbit is a Rust-based implementation of logistic regression and field aware factorization machines.
It runs on pure CPU and achieves extreme throughput compared to other similar tools. High throughput allows for a quick turn around in machine learning applied research. It can be used for hyperparameter optimization or feature and field selection and combination. It also has an on-line serving mode and is being actively used in production.
In this presentation, we’ll present how extreme speed and throughput were achieved - using Rust strong code specialization capabilities and careful benchmarking and code-tuning.

https://github.com/outbrain/fwumious_wabbit

-----
Paul Koerbitz
Paul is a software engineer with a PhD in financial mathematics. After spending ~10 years in various startups, he currently works for Google.

Lorenz Schmidt
Lorenz has just finished his master studies and is now pursuing a doctoral program. He worked for companies on NLP and speech enhancement systems.

Classical ML for Rust: linfa

Meet linfa - Rust’s take on classical machine learning in the spirit of scikit-learn. Linfa’s current maintainers will tell you about linfa’s ambitious goals, its current state and the next steps forward. We’ll also touch on how to implement classical ML algorithms in Rust and how you can be part of the journey.

-----
Daniel Mckenna
Daniel is a software engineer at Emotech, bringing machine learning to production with Rust. Before Emotech he spent 5 years working with C and C++ in aerospace and defence.

Multimodal Pronunciation Assessment in Rust

This talk will cover Emotech's Multimodal Pronunciation Assessment system which uses both audio and visual information to work out the quality of an English student's spoken pronunciation. This talk will cover some basic linguistics, the benefits of Rust, some of the existing ecosystem and the process of turning a python proof of concept into a production system.

https://github.com/xd009642