Aberdeen Data Meetup Sept 2018: MSc Data Science / AI Showcase
Description changed:
September's data meetup will be dedicated to highlighting the projects of students who have just completed their MSc Data Science and AI at Aberdeen Uni and RGU.
We'll have four short presentations of around 20 mins plus questions and loads of networking!
September's data meet-up is a showcase of projects of students who have just completed their MSc Data Science and AI at Aberdeen University and Robert Gordon University.
We've four short presentations, each of around 20 mins plus questions and loads of networking!
**************************
18:30 pizza and beer.
**************************
7pm - Talk 1. Susan Krueger - MSc Data Science, RGU:
"Visual Document Mining - making sense of seemingly unrelated collections of textual data."
**************************
8pm - Talk 3. Ian Watt - MSc Data Science, RGU:
"Blocklists, Bubbles and Cliques: identifying and measuring the effects of Twitter echo-chambers in the Scottish independence debate."
***************************
8.30 pm - Talk 4 (tbc) Aberdeen Uni
****************************
Abstracts:
Susan Kreuger:
In this talk, you'll find out what visual document mining is and how it can be used to make sense of vast and seemingly unrelated collections of text data. We'll look at a simple visual document mining tool in action and go under the hood to learn how it mines text and uses unsupervised machine learning to uncover hidden patterns and associations within the collection, which are then visualised for analyses and discovery by the user. My aim will be to unpack and explain two of the many approaches to machine learning for text data: topic modelling and the k-means clustering algorithm. If there's time, we may look at the Python code and tooling used in development as well as tips for storytelling with data.
----------------------------
Wenjun Liao (Claude):
Tackling fake news: how ensemble-based systems can outperform the state-of- the-art algorithms in news stance detection task with relatively simple architecture and easy implementation.
---------------------------
Ian Watt:
Using novel method to detect who actually uses a prominent distributed Twitter blocklist, and creating network graphs of 116 million users, we analyse whether or not using a blocklist exacerbates the echo-chamber effects which are prominent in social media.
*****************************
Thanks to MBN Solutions, Data Lab Scotland and Scotland IS for sponsoring this event.
Photo by Hal Gatewood on Unsplash
September's data meet-up is a showcase of projects of students who have just completed their MSc Data Science and AI at Aberdeen University and Robert Gordon University.
We've five short presentations, each of around 20 mins plus questions and loads of networking!
**************************
18:30 pizza and beer.
**************************
7pm - Talk 1. Susan Krueger - MSc Data Science, RGU:
"Visual Document Mining - making sense of seemingly unrelated collections of textual data."
***************************
7.25 pm - Talk 2. Wenjun Liao (Claude) - MSc Artificial Intelligence (AU)
"Using ensemble-based systems to tackle fake news."
**************************
7.50 pm - Talk 3. Ian Watt - MSc Data Science, RGU:
"Blocklists, Bubbles and Cliques: identifying and measuring the effects of Twitter echo-chambers in the Scottish independence debate."
***************************
8.15 pm - Talk 4. Chidozie Victor Nnachor, Aberdeen Uni
Database for Organic Rotation at Scotland's Rural College
****************************
8.40 - Talk 5. Miruna Clinciu, AU
Optimisation of oil and gas reservoirs by prediction of time-series data, using the SAX Algorithm and Long Short Term Memory(LSTM) neural networks.
Abstracts:
Susan Kreuger:
What is visual document mining and how it can be used to make sense of vast and seemingly unrelated collections of text data. We'll look at a simple visual document mining tool in action and go under the hood to learn how it mines text, using unsupervised machine learning to uncover hidden patterns and associations within the collection, which are then visualised for analyses and discovery by the user. We will explain two of the many approaches to machine learning for text data: topic modelling and the k-means clustering algorithm.
----------------------------
Wenjun Liao (Claude):
Tackling fake news: how ensemble-based systems can outperform the state-of- the-art algorithms in news stance detection task with relatively simple architecture and easy implementation.
---------------------------
Ian Watt:
Using novel method to detect who actually uses a prominent distributed Twitter blocklist, and creating network graphs of 116 million users, we analyse whether or not using a blocklist exacerbates the echo-chamber effects which are prominent in social media.
----------------------------
Chidozie Victor Nnachor:
The project is a centralised database system for Scotland's Rural college Cereal Yield field trials. The web application provides consistency and availability to the research data generated by the Aberdeen arm of the college. Other features include owner-initiated data uploads, User access hierarchy and fluid and automated audit trail of the Database. It also presents visualisation data for public consumption. It was developed by a team of 6 Msc Students at the University of Aberdeen.
----------------------------
Miruna Clinciu:
The oil and gas industry depends on physics based models to predict future performance of oil and gas reservoirs, with a large amount of data generated from well sensors.
With a greater uncertainty on historical data there is a need of exploring how we could use other data attributes collected at various points in our systems to close the gap, narrow uncertainties, and improve performance.
The water cut data is considered the key for optimization of the oil and gas reservoir and the forecasting of the point when the production should be shut down leads to increased profits.
The aim of this project is to predict the water cut, represented as time-series data, using the SAX Algorithm and Long Short Term Memory(LSTM) neural networks. For achieving an accurate forecasting model we need to fill the gaps by predicting the inferred data using different machine learning techniques.
*****************************
Thanks to MBN Solutions, Data Lab Scotland and Scotland IS for sponsoring this event.
Photo by Hal Gatewood on Unsplash
