Huang, students to improve cloud computing through EAGER grant
Professor Thomas Huang and a team of three students received a grant in late August from the National Science Foundation to research a more efficient learning algorithm search through cloud computing.
The grant is EAGER, Early-Concept Grant for Exploratory Research, which generally funds research in its early stages as a high risk/high reward project.
“The system we built may not work, but that’s a risk,” said Liangliang Cao, a student on Huang’s team. “However, this is also an opportunity to create a very efficient system, which is a high return.”
The team hopes to make cloud computing easier to use, cheaper and faster for numerical scientific applications.
They plan on doing this by building their own cloud computing system so they can control the environment. Then they will start testing ideas, starting with simple ones and progressively getting more complicated. By the end, they hope to have realized more real world applications.
“We have ideas we are going to try out,” Huang said. “The area of cloud computing is a very hot area. Everyone is talking about it.”
While cloud computing is already in use, processing high dimensional data, such as multimedia, is very slow.
Currently, several machines are required to process all the multimedia data -- which takes careful coordination between machines. Huang's team hopes to make that process faster so computers that are geographically far away from each other would be linked up to the same system.
“Images and video contain a huge amount of data and we need to deal with spaces of very high dimensions, so the current cloud computing architecture is not suited for this,” Huang said.
While they are conducting the research to better understand the cloud computing enviornment, there are real world applications, including enhanced online search functions, as well.
For example, if someone was uploading pictures to the internet, the computer could try and guess where the picture was taken based on comparing the image’s background to other backgrounds already on the Internet. Or if someone was watching a YouTube video and liked the shirt someone in the video was wearing, they could then find all the other videos on YouTube where that shirt appeared.
Cao is hoping that through cloud computing methodology, this can be done with greater speed and accuracy.
The research will be conducted at CSL, Beckman Institute, the Computer Science Department and NEC research labs.