JEDI workshops seek to overcome the traditional views of conference style presentations, by bringing together students and/or delegates to jointly participate in active research projects. While traditional conferences deliver a myriad of presentations, the JEDI, seeks to limit the amount of formal presentations in lieu of smaller working discussions. Students, researchers and professionals alike, work together on predefined projects in an attempt to either solve, or begin to solve the problems or questions posed. Projects are executed by teams that form spontaneously based on the particular interests of the participants (i.e. a participant would elect a specific project and then form a team with others who have elected that same project).
The upcoming era of Big Data promises to revolutionize scientific research. However, a new kind of scientist will be required to make the most of this opportunity. This workshop will train students in some of the techniques required to tackle Big Data Science problems. Projects spanning Agriculture, Astronomy and Health applications will allow the students to be exposed to applications in different realms and develop a data-driven approach toward scientific investigation.
The JEDI will take place in the lovely coastal town of Nosy Be, Madagascar from Friday, May 25th to Tuesday May 29th. This workshop will run in parallel with the SKA-driven Big Data Challenge in Africa: Science, Innovation and Opportunity Conference (http://www.idia.ac.za/conference/ska-driven-big-data) to be held at the same venue. JEDI participants will therefore have the unique opportunity of attending this conference at no extra charge and will be expected to stay for the remainder of the main conference as well. Students will thus arrive May 24th and depart June 1st.
A total of 20 students will be selected to attend the conference all expenses paid. The workshop aims to attract senior post graduate students (MSc or PhD), however any student with sufficient background in programming, particularly Python, is welcome to submit an application under the "Registration" tab by no later than 06 April 2018. Once the deadline has passed, applications will be reviewed and a final selection made. Successful students will be notified via email within 1 week of the closing date.
Astronomy : Radio Source Detection and Classification
Automated detection and classification of radio astronomy signals is becoming increasingly important with the dramatic growth of image data sets from next generation radio interferometer telescopes. In this project we will depoy and analyse algorithms to automatically detect and measure astronomical radio emission in images from data taken by SKA pathfinder radio telescopes. We will also develop analytical tools and statistical, intelligent approaches to identifying complex, multiple and component signals in the image data.
Health : Image-based Data Mining to Improve Radiotherapy for Cancer Treatment
Radiotherapy is a treatment for cancer in which radiation is directed into the patient to kill cancerous cells. Unfortunately sometimes the treatment does not work. In this project we will use data from several radiotherapy treatments to try and correlate the radiation dose with, for example, survival/recurrence aiming to improve radiotherapy for future patients. The project will follow the whole process of image-based data mining, including the use of tools to perform image registration and analysis techniques to interrogate the data.
Astronomy : Machine Learning for Extracting Stellar Rotation Periods from Kepler data
Stars spin – and the rate at which they spin has a profound effect on their health. So measuring the rotation of different types of stars is an important scientific challenge. Fortunately stellar rotation is something that can even be measured directly from time dependent brightness: stars don’t have smooth surfaces, like our own Sun they are spotted with activity. When we observe them, this non-uniformity is reflected in the data as a quasi-periodic variation in their brightness. However, the time dependent variability in the emission from spotted, rotating stars is often not perfectly periodic. Spots on the surface of stars have a tendency to move and disappear or appear. This means that fitting a straight forward sinusoidal model to the data is not very reliable – and fitting a full physical model is extremely complicated. This project will investigate machine learning approaches, starting with Gaussian Process Modelling, to extract stellar rotation periods in an automated way using data from the Kepler satellite.
Agriculture : Optical and Radar Imagery for Sustainable Agriculture
Sustainable development of the world’s food production and supply systems is essential for meeting the demands of an increasing global population. New and more accessible satellite imaging technology has huge potential for improving agricultural efficiency. Utilising the latest earth observation imaging platforms (ESA’s Sentinel missions and NASA’s LandSat) we will look at a range of uses and techniques for optical and radar imaging in agriculture.
Astronomy : Building a Workflow to Deploy a Radio Astronomy Calibration & Imaging Pipeline to the Cloud
Reproducible scientific pipelines have historically been a challenge in most computationally focussed fields. The Common Workflow Language (CWL) is a standard for specifying and simplifying the process of executing scientific processing pipelines, while ensuring simplicity of use, portability and reproducibility. This project will focus on transforming an existing Python and Jupyter Notebook based pipeline for calibration and imaging of radio telescope data. Participants will complete a CWL-ified workflow in a Jupyter environment and execute it on the IDIA cloud.
This workshop has been geared toward students who have a solid background in coding and information technology with a strong emphasis on analysing astronomical data. Two additional projects will however be offered on both medical imagery and graphical information systems (detailed descriptions can be found under the "About" tab).
The meeting will run for 5 days and will be held on the lovely coastal island of Nosy Be, showcasing the beauty and culture of Madagascar. The planned venue is the Ravintsara Wellness Hotel. Madagascar is an island located in the southern part of Africa, between the Indian Ocean and the Mozambique channel. The latitudinal and longitudinal extent of the country is 20° South latitude and 47° East longitude, with a total area of 587 040 km2. Nosy Be is 3 hours and 30 minutes flight away from Johannesburg. For more information on tourism within Nosy Be please visit http://www.nosybe-tourisme.com/ or https://nosybe-tourism.com/en/practical-informations/for some practical information. If you require any information on power sockets and adapters suitable for Madagascan outlets, please visit: https://www.power-plugs-sockets.com/madagascar/. Local cabs/taxies are readily available at the airport and cost approximately R 200.
Successful candidates will be prodived with sharing accommodation (bungalows) for the duration of their stay at the wonderful Ravintsara Wellness Hotel. Bungalows can accommodate 3 to 4 people each. For more information or images of their facilities, please visit http://www.ravintsarahotel.com/.