Private:progress-abdelsadek

Spring 2014 (RA)

• Courses:
  • none

Jan 10th

• Finished camera-ready version of mmsys14 paper.

Fall2013 (RA)

• Courses:
  • CMPT-726: Machine Learning

Aug 30th

• First Draft Report

Summer 2013 (TA)

• Courses:
  • CMPT-880: Large-scale Multimedia Systems and Cloud Computing

July 29rd

• Gave the project presentation "Large-Scale Content-Based Image Retrieval"
  • Full slides can be found here.
• Did some elementary experiments on image search task
  • Using around 20,000 images having around 20,000,000 feature points.
  • Using Amazon EMR, experiments on 4 and 8 dual core machines.
• Finished the project report, including the full description of the previous experiments, and their plots.
  • Full report can be found here.

July 23rd

• Implementing image downloader, to download some datasets from the web.
• Working on the course project codes, and datasets.
• Writing the course project report.

July 9th

• Enhanced and optimized the old KD-tree for better performance.
• Implemented an efficient version of PCA. -> a bit slower at building, but increased the core KNN accuracy by 15% to 25%
• Running Experiment set 1 mentioned earlier in Evaluation setup report, and plotted its results

July 3rd

• Evaluation setup – Experiments design

Designed some experiments sets to evaluation both accuracy and performance of the system in both general usage case, and for the 2 target applications. Evaluation setup report can be found here.

• Implementation
  • Implement image matching algorithm for matching similar images from multiple similar local features. Used ranking method instead of absolute distances
  • Modified the code to make it more general and made some memory optimizations

• Refined and Enhanced design
  • Added map/reduce job for constructing the index bins in a distributed way, instead of the centralized way in the previous design
  • Element the bottlenecks, the designed now can handle even larger reference datasets then the previous one.

June 25

• Prepared and gave presentation on Map/Reduce and Hadoop

Full slides can be found here.

• Working on the Proposed Solution chapter in the thesis, added more details, still some more refinements to do.
• Working on enhancing the implementation of the index and optimizing the code, and make it more general
  • stuck a little bit because I didn't have development environment nor AWS account, Now I installed Hadoop virtual cluster on my local machine, Hopefully will have some experiments by next week.

June 11

• Wrote a respond report for the patent reviews.

Full report can be found here.

• Revised and modified the last version of the index design.

Full design document can be found here.

• Created a thesis folder, and started adding up to it.

Full Thesis can be found here.

May 31

• Review what already done, applications for my work, similar works, ways of evaluation, and possible extensions.

Full report can be found here.

Spring 2013 (RA)

• Courses:
  • CMPT-820: Multimedia Systems

Feb 28-Mar 14

• working on figuring out the rest of implementation details for Distributed LSH over Map/Reduce.
• had a meeting with Yahia for getting the last version of the code, sync with the get hub, and the new design of the system.
• setting up and testing the development environment on my local machine.

Feb 14-Feb 28

• I did a report giving a quick overview on Locality-Sensitive Hashing for solving near neighbors search problem. showing its main ideas, algorithms, and techniques. also introducing some early thoughts how to implement it in our distributed index and showing some of the issues/challenges of its implementation.

Full report can be found here.

Feb 5-Feb 14

• Reading and Surveying about LSH
• Did some initial ideas about LSH and how to plug it into our system.
• working and discussing with Dr. Maysam about he idea and implementation for the index.

Jan 22-Feb 4

• Working on evaluation setup for the index.

Fall 2012 (RA)

• Courses:
  • CMPT-705: Design/Analysis Algorithms
  • CMPT-741: Data Mining

• KNN

Did a survey on KNN algorithms and techniques.

Full survey can be found here.

• Industrial Automation as a Service
  • Doing Dr. Tamir Experiments