Deep Dive into Deep Learning

I was privileged and excited to receive an entrance scholarship from Healthy Brains for Healthy Lives to the IVADO / MILA Deep Learning Summer School, which took place this past week at UQAM. The week-long school provided a deep dive into machine learning and deep learning techniques and had us thinking about the best ways to use these methods within our own big datasets. All of the talks were aimed at a broad audience, with theoretical lectures first thing in the morning, mathematical ones at midday, and computational tutorials later in the afternoon, which really helped in understanding the methods across different perspectives.

I particularly appreciated Jeremy Pinto’s lectures on CNNs since it has such overlap with my current work on predicting changes in attentional and mind wandering states. Khimya Khetarpal also had a great talk on RL that touched upon some of my interest in predicting eye movements within real-world environments, and it was quite a callback to when I learned about Markov models back in 2008 during my time at Queen’s! I’d also be remiss to not mention Golnoosh Farnadi’s talk on how to account for bias and discrimination in AI, which was a great bookend to finish up our Friday afternoon and to think about the idea that sometimes the most efficient way that machines “learn” is by exploiting inefficiencies and loopholes that already exist in the system.

Beyond being incredibly informative and well-organized (a huge thanks to Nathalie Sanon and Nadia Bennaïli on keeping things running smoothly all week), I was personally grateful that IVADO and MILA were committed to reducing plastic consumption during the week-long school, since these events can usually leave a huge footprint in terms of waste – all of our lunch and break materials were 100% biodegradable, reusable, and compostable, and they had a system set up so we could recycle our name tags and any other materials we were given once the summer school was done. Really loved seeing these institutes care about their impact beyond the learning material!

I’d highly recommend this summer school for those interested in getting started with ML and DL techniques (Vancouver-ites, they’ll be running this school at UBC soon!), and I can’t wait to start training some neural nets myself! Stay tuned to my Software page for some implementable examples and cool visuals to follow!