Introduction to Neural Networks and Deep Learning

Neural Networks are a very effective machine learning tool for building predictive models from big data. We will discuss applied examples of problems that can be solved by these methodologies, emphasizing the high-level inner-working concepts behind these methodologies. You will learn about:

• Perceptron: A foundational building block of Neural Networks
• Basic Neural Network Models: uncovering architecture essentials
• Classic Deep Neural Network (DNN) models: Intro to deep learning
• Convolutional DNNs: image processing
• Recurrent DNNs: processing sequences (e.g., words)
• Long Short-Term Memory DNNs: demystify the memory

This is the 3rd webinar in a 3-part series focused on the machine learning.Other workshops in this series:

• Oct 13, Introduction to Machine Learning
• Oct 20, Introduction to Predictive Performance Assessment

About the Machine Learning Series

This webinar series was made possible via joint collaboration between Faculty of Science (FOS), Applied Statistics and Data Science Group (ASDa) and Graduate Pathways for Success Program (GPS). It was originally funded by the UBC Science Strategic Innovation Fund and is now currently funded by the Graduate Pathways for Success Program. Don’t let data overwhelm you! Join us for this machine learning webinar series and use our expert guidance to empower yourself with deeper understanding of data!

Why attend?
Machine learning and Statistics are dynamic scientific disciplines that enable reaching meaningful insights from data, it’s not just about numbers. To produce reliable results, you need to justify the choice of the applied statistical or machine learning methods and models, as well as validate the underlying assumptions.

What to expect?
This webinar series provides introductions to the foundational concepts of advanced statistical methodologies in machine learning algorithms. We will provide practical insights by discussing different machine learning methods, their appropriate application and how to assess their predictive performance. The aim is to equip you with a deeper understanding of the key concepts of statistical and machine learning methodology, rather than solving specific project problems. While R code for hands-on guidance may be provided, the emphasis of this series will be on the methodological aspects. Each webinar is a self-contained introduction, but as topics become increasingly complex with each consecutive webinar, some aspects will be built on concepts taught in the previous sessions. Hence, there is benefit in attending all the webinars in this series.

ABOUT ASDa

The Applied Statistics and Data Science Group (ASDa) in the UBC Department of Statistics participates in collaborative research and provides statistical consulting services. ASDa expertise includes problem formulation, translation of research questions into testable statistical hypotheses, design of experiments and sampling plans for surveys, the choice and explanation of statistical methodology, statistical computing and graphics, the interpretation of findings and more. ASDa also plays an active role in continuing statistical education on and off the UBC campus, giving seminars, webinars, hands-on workshops and courses on statistical concepts and methodologies to various departments, research groups and at teaching hospitals. If you are a graduate student and have questions about your specific project, please see this website to book a one-hour free statistical consultation.

Facilitator

Biljana Jonoska Stojkova, PhD is a Senior Statistical Consultant with ASDa, participating in collaborative research, providing statistical consulting services and tailored education on statistical concepts and analytics tools on and off campus. Biljana is particularly passionate about developing statistical education and training programs for graduate and undergraduate students. She completed her PhD in Statistics at SFU in 2017, where she focused on developing Bayesian algorithms and methods for multi-modal posterior parameter distributions, which were applied to differential equation models, mixture Gaussian models, epidemiological and ecological models. In the previous roles she has gained experience with probabilistic models to determine different patterns of user behaviour from chat messages, with development of relational databases and with machine learning algorithms such as supervised and unsupervised learning. In her consulting role Biljana continues to strengthen her skills in problem formulation, study design, grant proposal development, analysis and implementation, preparation of scientific manuscripts and continued education of non-statisticians on and off the UBC campus, giving webinars, workshops and courses on statistical concepts and methodology to various departments, research groups and at teaching hospitals.