- Greater Pittsburgh Area, US PA
Sep 2010 - Jun 2014
Siddaganga Institute of Technology, Tumkur, INDIA
Image segmentation to improve prediction of terrain traversibility using visual and/or thermal data
May 2015 - Aug 2015
- Under the guidance of Prof. WIlliam 'Red' Whittaker
- Used machine learning algorithms like random forests and support vector machines on thermal and visual imagery to pick out different classes in the images, such as rocks, soil, vegetation, machinery etc.
- Worked with a pixel-by-pixel classification using Texturecam and random forests; switched over to a superpixel implementation, by growing homogeneous regions.
Scene recognition using Spatial Pyramid Matching
Sep 2015 - Sep 2015
- Used computer vision techniques to build an algorithm to match an image to its correct scene.
- Used bag-of-words approach and formed visual words out of images in training set, k-means for clustering, and spatial pyramid of 3 levels to match scene
Movie recommender system using collaborative filtering and SVD
Mar 2015 - May 2015
- Made use of the MovieLens dataset to achieve this task.
- Used collaborative filtering as a first initial step, with Pearson correlation as a measure of distance between users/movies. Correlations based on users, and movies separately.
- Singular Value Decomposition to (successfully) increase accuracy.
Biological Expression Language parsing and extracting meaningful data from biological papers
May 2015 - Aug 2015
- The tool was developed using Python. Currently, it reads a file of BEL statements, parses each statement into its components, compares each statement against every other statement based on the parsed components, and assigns a similarity score to each pair of statements.
- All the similarity scores are tabulated into a correlation matrix for better understanding.
An experiment : 'Can robots invoke empathy in humans?'
Jan 2015 - May 2015
- Devised and implemented a social experiment to test on human participants to see if robots can effect humans' empathy towards them.
- This involved getting the participant to interact with the robot under controlled conditions, while monitoring the situation.
- Data regarding the test and control subjects was collected and analyzed to help draw a conclusion.
Have developed all machine learning and vision related projects on Matlab
Worked with Python over summer developing tool for parsing syntactical language
Have taken an introductory course in computer system; coded the malloc function, implemented a tiny shell, and a cache as projects for the corurse