From Chaos to Productivity: Organizing Your Machine Learning Project

Title:
From Chaos to Productivity: Organizing Your Machine Learning Project

Abstract:
Machine learning practitioners need to iterate through many models very quickly, however, there is no standard approach for this process. If properly designed, such a process should allow the researcher to train dozens of model types with various combinations of hyperparameters using any number of training datasets. The researcher should be able to keep everything organized, quickly evaluate the results, and use scientific principles to figure out the right direction to find that one golden model. Moreover, the machine learning practitioner must also deal with various deep learning frameworks, repositories, and data libraries, as well as deal with hardware challenges involving GPUs and cloud computers. In this talk, we present a guide to organizing your projects and to help you create a true laboratory for model exploration that will hopefully motivate you to become a better machine learning practitioner.

Speakers: Aleksandr Gontcharov and Jason Carayanniotis
Affiliation: Machine Learning Engineers at IMRSV Data Labs

Bios:
Aleksandr Gontcharov is a machine learning engineer at IMRSV Data Labs specializing in developing deep learning algorithms for projects ranging from natural language processing for evidence-based medicine to audio processing and computer vision. Previously, he was a data scientist at the Government of Canada where he championed the exploration of AI as a tool to automate workflow. His educational background includes a Master of Science in Mathematics from the University of Ottawa and an Honours Bachelor of Science in Mathematics from the University of Toronto.

Jason Carayanniotis is a machine learning engineer IMRSV Data Labs with a background in mathematics and physics from Memorial University of Newfoundland. His current projects are in the Natural Language Processing and Generative Adversarial Networks. His previous work includes championing the exploration of Deep Learning to design holographic displays by exploring the properties of nanophotonic structures and performing physics simulations on compute clusters at Avalon Holographics.

~ Light refreshments will be provided ~

Agenda:
6:30 Arrival & Networking
7:10 Approx start of the talk
8:00-8:30 Post talk networking & discussion