Featured Job

Sign In
 [New User? Sign Up]
Mobile Version

Machine Learning for Earth Science (VIP) - NASA

NASA - National Aeronautics and Space Administration


Location:
Moffett, Oklahoma
Date:
03/02/2017
Categories:
  • Data Analytics
  • Data Scientist
  • Business Intelligence
NASA - National Aeronautics and Space Administration
  •  
  • Save Ad
  • Email Friend
  • Print
  • Research Salary

Job Details

This work is in support of the Data Science group's goal of modeling physical systems using a variety of machine learning techniques. Of particular interest is the technique of deep learning and its use in unsupervised learning/modeling. To accomplish this, work requires familiarity with techniques for efficient handling of "big data" i.e., data with large volume, high velocity, high variability, and changing definitions. The main steps of the task would be Familiarize oneself with the different open source platforms for deep learning and choose the best one for task at hand so that recoding existing modules is minimized. Familiarize oneself with the data set(s) at hand Code appropriate deep learning algorithm for the modeling task Optimize code for performance efficiency Test code on existing data sets Compare results with other baseline methods such as symbolic regression

Requirements

Requirements

Familiarity with large-scale data mining and machine learning techniques for clustering, classification, regression, and anomaly detection. Working knowledge of neural networks and deep learning methods. Programming in C/C++/Matlab/java/python. Familiarity with one or more open source platforms for deep learning such as mxnet, Theano, or TensorFlow Ability to apply a computer cluster for parallel computing projects such as MPI, Hadoop, Spark, or a PBS scheduler. Demonstrated ability to work in small teams. Excellent oral and written communication skills Excellent references Should be enrolled full time or part time for undergraduate/graduate studies in a relevant discipline such as Computer Science, Electrical Engineering, Statistics. Should be eligible to work legally at least 20 hrs a week onsite.

 

Expected Opportunity Outcome

  • Software prototype for the desired task -Technical report or submission draft for a conference paper depending on the contributions