We are delighted to announce that we will be curating a Machine Learning Bash on Monday 23rd April in the Black Box, co-sponsored by Liberty IT and Instil. We’ve lined up four great speakers for you - see the descriptions below for full details.
The talks will be divided into two groups, with a break for food and beverages in between. Catering will be provided courtesy of our sponsors (details still be to confirmed but bring an appetite).
---------------------- Talk Descriptions ----------------------
Betting on Artificial Intelligence: Three Use Cases to Replace a Job (Jason Bell)
Jason takes us through a whistle stop tour of three popular machine learning algorithms. Plus he'll replace someone's job on the night (he won't say who's yet) and get the audience to bet on the algorithm that will replace them. Which algo will win?"
Natural Language (mis)Understanding (Gillian Armstrong)
As a developer I spend a lot of time being cross that a computer did what asked it to, and not what I meant it to. With all the AI services out there allowing Natural Language Processing and Understanding, surely the time has come when computers can finally really understand us? Let’s take a look at how easy it is to build a chatbot using an NLU service, and how easy it is to break it if you haven’t thought through your interaction model design and training data carefully. We may need other humans for a while yet…
The Potential of Streaming Data Analytics in the ICU (Charles Gillan)
Intensive care units (ICU) routinely collect vast volumes of measurements of physiological data. A patient in a modern ICU is surrounded by a large number of monitoring devices, yet the vast majority of the data is either dropped or stored and forgotten. Recently the viability of monitoring physiological parameters to detect sleep apnea in neo-natal ICU has been proven by McGregor et al in Canada. In our work at QUB, we have focused on analysis of parameters associated with respiratory ventilation. The talk will briefly explain characteristics of lung ventilation and then focus on the design and implementation of our software, the VILI Alert System. We will present preliminary results and explain how we are extending the research.
HODL-AI: Live Coding RNN on the Blockchain (Andrew Bolster)
Fully aware that Live Demos are the worst idea ever, Bolster goes-a-wandering through the collection, analysis and statistical characterisation of a few choice Blockchain Cryptocurrencies valuations, using Recurrent Neural Networks to model their behaviour and identify what strategies could have historically given the most ROI. It's bound to be a disaster but at least it'll be hilarious.