You are viewing content from a past/completed QCon

Scaling & Optimising the Training of Predictive Models

Modern Machine Learning has brought with it countless advances, both algorithmically and with respect to tooling; there is relentless growth on all fronts. Nevertheless, we are faced with a multitude of challenges when trying to pull all these threads of progress together in a meaningful way, especially when operating on an industrial scale. With huge volumes of data, models of increasing complexity, and a list of hyper-parameters as long as your arm, writing a big for-loop, and using a big workstation is rarely sufficient.  

This talk presents the core building blocks of an entire tool-chain, which is able to deliver on all the above challenges... and more! We assume a "living" model already exists, i.e. a model ready to be trained, tweaked and optimized as efficiently as possible, on a growing dataset. It is not a thought experiment but based on a real solution that was put into production earlier this year.


Nicholas Mitchell

Machine Learning Engineer at @argoai

Nicholas is a Machine Learning Engineer at Argo AI, solving the task of perception for autonomous vehicles. He is also a part-time researcher, completing a Ph.D. in cross-domain applications of AI.


Tuesday Nov 10 / 11:40AM PST (40 minutes )

TRACK Machine Learning for the Software Engineer ADD TO CALENDAR Add to calendar

From the same track

View full Schedule

3 weeks of live software engineering content designed around your schedule.

Don’t miss out! Save your seat now