|Position Title||Deep Learning of Large Biomedical Data using GPGPU|
|Summary||In this projects. two or more undergraduate students will work as a team, while utilizing the XK7 compute nodes to design/implement machine learning algorithms on large biomedical data sets. In the process, the students will acquire knowledge in machine learning/deep learning, parallel computing and HPC related subjects.|
|Job Description||The interns should have already learned programming skills in C/C++, as well as been familiar with working in Linux/Batch environment. They also need to have proper background in Maths, including linear algebra, statistics.|
Throughout the course of a year, the undergraduate students are expected to learn and profess following skills:
1) Basic general knowledge in HPC, including parallel programming skills in MPI/OpenMP/CUDA;
2) Basic general knowledge in Datamining, including clustering, classification, association and anomaly detection;
3) Theories and practice in Artificial Neural Networks, both in terms of basic research and in applied computing.
Through out the year, and especially in the second half of the year, the student interns should
1) read current literature in deep learning,
2) implement and benchmark known algorithms using the Blue Waters,
3) propose innovative algorithms to process specific biomedical data sets using NVIDIA GPGPU available at Blue Waters.
A faculty in the Computer Science Department, as well as advisors with bio-research background will guide the students in this study. We have some specific data sets in mind, but would also work on general problems to give the students a rich learning experience.
|Conditions/Qualifications||In addition to the two students already identified, if other students want to join, the undergraduate interns should have already processed programming skills in C/C++/Python/R, as well as familiar with working in Linux/Batch environment. They also need to have proper background in Maths, including linear algebra. They should live within a 50 miles range (which includes the Baltimore and D.C. metro), and commit to weekly meetings in the summer, and monthly meetings in the rest of the year.|
|Location||Hood College of Frederick, Maryland|