Itauma came up with this project idea while exploring research ideas relating to the classification of large scale image datasets which was the subject of his computer science thesis. The objective of the research is to explore a Big Data analytic platform that can scale to the size of the data and is suitable for implementing machine learning algorithms. While carrying out this research, he discovered HPCC Systems which caused him to spend almost 5 months in 2015 working on his ideas using our open source platform.
This project required feature representation from face images and sound data and classification based on these representations. The three main aims of the project were:
- To formulate a new feature learning and recognition framework using a multimodal strategy, leading to high classification accuracy.
- To use HPCC Systems to implement the feature learning and object classification tasks.
- To use HPCC Systems to speed up the run time of any computationally intensive algorithm
This work shows that HPCC Systems can successfully be used for image analysis. Itauma presented his project findings at the International Conference for Machine Learning and Applications (ICMLA) and introduced the HPCC Systems platform to the ICMLA community. Read his paper for more in depth information about this project.
While completing his Masters program and as a member of the Data Science and Analytics Laboratory at Wayne State University, Itauma made more progress on this project. He has completed work to optimize the unsupervised learning algorithms used for feature learning, resulting in an improvement on the classification accuracy on new images.
Item is continuing with his studies at Wayne State University and intends to continue to explore the use of HPCC Systems for other applications such as text analytics, learning analytics and education analytics to improve personalized learning.
Itauma is a great advocate for the HPCC Systems open source platform and was presented with a Community Ambassador Award at our Engineering Summit in 2016 as a token of our thanks and appreciation.
Itauma's poster presentation entered into our competition held on Community Day at the HPCC Systems Engineering Summit in 2016.