Automatic Machine Learning and Open-World Visual Recognition

We have a venue in Central Dublin. Please note that spaces are limited and entry is on first come first in basis.

=======================================================

Agenda:

6:30 - 6:45 - Doors open + Networking

6:45 - 7:30 - First Talk + Q&A

7:30 - 7:40 - Short Break

7:40 - 8:25 - Second Talk + Q&A

=======================================================

First Talk

Title: Scalable Automatic Machine Learning with H2O’s AutoML

Abstract:

In recent years, the demand for machine learning experts has outpaced the supply, despite the surge of people entering the field. To address this gap, there have been big strides in the development of user-friendly machine learning software that can be used by non-experts. The first steps toward simplifying machine learning involved developing simple, unified interfaces to a variety of machine learning algorithms (e.g. H2O).

Although H2O has made it easy for non-experts to experiment with machine learning, there is still a fair bit of knowledge and background in data science that is required to produce high-performing machine learning models. Deep Neural Networks in particular are notoriously difficult for a non-expert to tune properly. In order for machine learning software to truly be accessible to non-experts, H2O has designed an easy-to-use interface which automates the process of training a large selection of candidate models. H2O’s AutoML can also be a helpful tool for the advanced user, by providing a simple wrapper function that performs a large number of modeling-related tasks that would typically require many lines of code, and by freeing up their time to focus on other aspects of the data science pipeline tasks such as data-preprocessing, feature engineering and model deployment.

H2O’s AutoML can be used for automating the machine learning workflow, which includes automatic training and tuning of many models within a user-specified time-limit. The user can also use a performance metric-based stopping criterion for the AutoML process rather than a specific time constraint. Stacked Ensembles will be automatically trained on the collection individual models to produce a highly predictive ensemble model which, in most cases, will be the top performing model in the AutoML Leaderboard.

In this talk, Joe will explain the APIs (R, Python and H2O Flow) for AutoML and then go through some use case examples.

Bio:

Jo-fai (or Joe) is a data scientist at H2O.ai. Before joining H2O, he was in the business intelligence team at Virgin Media in UK where he developed data products to enable quick and smart business decisions. He also worked remotely for Domino Data Lab in the US as a data science evangelist promoting products via blogging and giving talks at meetups. Joe has a background in water engineering. Before his data science journey, he was an EngD research engineer at STREAM Industrial Doctorate Centre working on machine learning techniques for drainage design optimization. Prior to that, he was an asset management consultant specialized in data mining and constrained optimization for the utilities sector in the UK and abroad. He also holds an MSc in Environmental Management and a BEng in Civil Engineering.

Second Talk

Title: Expecting the Unexpected: Open-World Visual Recognition.

Abstract:

In a real-world setting, visual recognition systems will almost certainly encounter inputs that were not anticipated when the system was built. For example, what happens when a system that was trained to differentiate between cats and dogs is deployed in the wild and encounters an elephant?

We solved this problem by integrating information from knowledge graphs into a visual recognitions system. Knowledge graphs are a rich source of structured information. They are thus a natural choice for extracting semantic meaning about concepts. A recognition system can subsequently be built on top of this information. We can then make more interpretable predictions in the form of properties rather than class labels.

This work can be applied in security, assistive technology, drone-automation and smart cars.

Bio:

Vincent is a researcher at IBM Research Ireland. His work aims to push the boundaries of what is possible with AI. His recent work focusses on using deep learning for visual recognition and scene understanding in open-world setting.

He also specializes in applying machine learning models to physical systems including renewable energy, transportation, and atmospheric dynamics. He has worked on multiple client engagements including an grid-forecasting project in vermont, recently nominated for the Royal Irish Academy AmCham innovation award.

Vincent obtained his PhD in physics and finance from the University of Arizona in 2011. He has conducted research at Princeton University in NJ, CNRS in France, and Leiden University in the Netherlands.

http://researcher.ibm.com/researcher/view.php?person=ie-VINCENTL