Free

Colloquium: Applications of Deep Learning

Event Information

Share this event

Date and Time

Location

Location

University College Dublin

Stillorgan Road

O'Brien Centre, Moore Auditorium

Dublin

Ireland

View Map

Friends Who Are Going
Event description

Description

Why attend:

Deep Learning (DL) has recently found an impressive number of applications ranging from self-driving cars to medical diagnosis, from machine translation to virtual reality. The recent success of DL is in part due to our ability to utilize large amounts of data, and in part due to novel model architectures that tackle different types of problems.

Sustainable growth of the value brought by DL will depend on continued investment in new architectures and training strategies, as well as a strong knowledge of the subject fields to which we seek to apply these tools.

This workshop covers both sides of this challenge in a three session format.

1) Human Centric Applications
2) Applications to Natural Images
3) Fundamentals of Deep Learning

Program:

  • 09.30 Registration
  • 09:40 Dr. Wendy Belluomini, IBM Research - Ireland and Professor Orla Feely VP Research UCD: Welcome and Introduction


Session 1) Human Centric Applications

  • 10.00 Dr. Debasis Ganguly, Research Scientist, IBM Research Title: "Extracting decision elements from meetings with a Deep Learning based approach"
  • 10.30 Dr. Brian Mac Namee, lecturer in the School of Computer Science at University College Dublin Title: "Putting the Human Deep Into the Machine Learning Loop"
  • 11.00 Coffee Break


Session 2) Applications to Natural Images

  • 11.20 Dr. Eric Risser, CTO at Artomatix Title: "Introducing a Stable and Controllable Parametric Model for Neural Texture Synthesis, Time-Varying Weathering and Style Transfer."
  • 12.00 Dr. Vincent Lonij, Research Scientist, IBM Research Title: "Expecting the Unexpected: Open World Visual Recognition"
  • 12.30 Lunch


Session 3) Fundamentals of Deep Learning

  • 13.30 Dr. Bowen Zhou, Director, AI Foundations, IBM Research Title: TBD
  • 13.50 Ruihai Dong, Lecturer School of Computer Science at University College Dublin Title: Exploitation of Deep Learning for Recommender Systems
  • 14.10 Dr. Barak A. Pearlmutter, professor in the Hamilton Institute at NUI Maynooth Title: "Automatic Differentiation from Vacuum Tubes to Big Data"
  • 14.45 Discussion
  • 15.15 Close / post workshop




Speaker Biographies:

Debasis Ganguly

Debasis Ganguly is a researcher at IBM Research Labs, Dublin, Ireland. He completed his PhD and postdoctoral research from the ADAPT centre, Dublin City University.

His research interests primarily include incorporating term relationships, including the use of topic models and word/document vector embeddings, into retrieval and feedback models for improving search effectiveness. He is also interested in cross-language and multilingual document search, a particular example of which is the case of retrieving relevant information from a collection of code mixed documents.

Dr. Brian Mac Namee


Dr Brian Mac Namee is a lecturer in the School of Computer Science at University College Dublin; a principal investigator at the Centre for Applied Data Analytics Research (www.CeADAR.ie); and a funded researcher at the Insight Centre for Data Analytics (www.insight-centre.org). Previously, he has lectured at the Dublin Institute of Technology and Trinity College Dublin. Brian is also the Director of Training for The Analytics Store, a data analytics consultancy and training company. In 2015 Brian co-authored the textbook "Fundamentals of Machine Learning for Predictive Data Analytics: Algorithms, worked Examples and Case Studies" which was published with MIT Press (www.machinelearningbook.com). Brian's research focus is on the intersections of data analytics, machine learning and human data interaction, and he leads a team of 10 researchers working in these areas at UCD.

Eric Risser



Dr. Eric Risser is an expert in the combined fields of artificial intelligence and computer graphics and is a pioneer in machine creativity. Before founding Artomatix, he authored six technical publications during his academic career at Columbia University and Trinity College Dublin. He has given talks at top industry/academic conferences such as Game Developers Conference (GDC) and Siggraph.

Talk Description: This talk introduces a powerful new parametric model for optimization based texture synthesis using neural networks. This texture synthesis approach can be extended to also perform time-varying weathering and style transfer. This new parametric model utilizes histograms and co-variance of neural activations to achieve state-of-the-art results on all three applications. A multiscale coarse-to-fine synthesis strategy is also introduced, which improves both synthesis quality and speed.

Vincent Lonij

Vincent Lonij 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.


Dr. Barak A. Pearlmutter


Dr. Barak A. Pearlmutter is a professor in the Hamilton Institute at NUI Maynooth in Co. Kildare, Ireland, with a primary appointment in the Department of Computer Science. He is the principal of the Brain and Computation Lab there. He was a professor in the Department of Computer Science at the University of New Mexico in Albuquerque, with adjunct appointments in the Department of Neurosciences and the Department of Electrical and Computer Engineering.


Dr. Bowen Zhou

Dr. Bowen Zhou is the Director of Artificial Intelligence Foundations (AIF), the first horizontal centralized machine/deep learning and AI research Lab at IBM Research that he founded in January 2017. AIF's mission is to democratize AI by improving AI’s learning capability and time-to-value. The research areas include foundational learning paradigms for AI, new architectures and algorithms for deep learning and end-to-end learnable AI. On its extended research team, AIF is founded with IBM's best talents and minds who were behind a number of major AI and learning milestones in history and who are making ongoing exciting new progresses, ranging from Deep Blue, TD-Gammon (the first self-teaching neural network that learned to play backgammon at human world championship level), IBM's Jeopardy!, to speech, vision, and natural language understanding etc, to name a few.

Dr. Ruihai Dong is a lecturer in the School of Computer Science and INSIGHT Centre UCD. Ruihai’s background is in AI field and Recommender Systems. His primary research is focused on developing techniques for dealing with user generated content, mining opinions and sentiment from such content for the purpose of building next-generation Recommender Systems.



Share with friends

Date and Time

Location

University College Dublin

Stillorgan Road

O'Brien Centre, Moore Auditorium

Dublin

Ireland

View Map

Save This Event

Event Saved