Sei sulla pagina 1di 2

RUSHABH

AGRAWAL

MALE, 22 YRS

LINKEDIN

WEBSITE

EDUCATION
B.E. Computer Science, MSc. Mathematics
(dual degree)

2008-2013 B.I.T.S. Pilani, Pilani Campus

8.14/10

HSC (Class XII)

2008

Sathaye College

93.83% aggregate

ISC (Class X)

2006

Lokhandwala Foundation School, Mumbai

93.33% aggregate

Computer Science Courses : Data Structures and Algorithms, Operating Systems, Theory of Computation, Database Systems,
Programming Language and Compiler Construction, Computer Networks
Mathematics Courses : Algebra, Topology, Functional Analysis, Differential Geometry, Measure and Integration, Elementary Real
Analysis & Group Theory
Statistics Courses : Probability & Statistics, Optimization, Operations Research,, Graphs and Networks, Numerical Analysis
Electives Completed: Machine Learning, Data Mining, Pattern Recognition, Object Oriented Programming, Fuzzy Logic
Research Interest: Machine Learning and Data Mining. Besides courses, have also done a lot of self-reading. Also took the
initiative to form a Machine Learning-Student Study Group, which meets weekly to discuss papers and various ML Algorithms
RESEARCH EXPERIENCE <Link>
Central Electronics Engineering Research Institute (CEERI), Pilani
Under Dr. A.S. Mandal (Scientist-F, Head, Cognition and Perception Lab)
Worked along with doing course-work from Sep, 2010 to May 2011 and as summer intern, from May 2011-July 2011
Worked on an ongoing project of Facial Emotion Recognition.
Also tackled the NP-hard problem of Steiner Minimal Tree using self-organization techniques (subsequently published)
Link Prediction (LP) in large scale Social Networks (current Thesis, from Aug-2012 to Dec- 2012)
Under Prof Sundar S Balasubramaniam, BITS Pilani
Abstract : Structure Preserving Embedding algorithm (by Blake et.al., ICML-2009) preserves the global topological
properties of a graph while embedding it in a low-dimensional space. SPE when used with algorithms like Minimum
Volume Embedding (SPE+MVE) performs very well as a dimensionality reduction algorithm. This could be used to reduce
size of Social Network Graph Matrix before learning a model for Link Prediction in the large Online Social Networks. The
only drawback is the large computational complexity of SPE+MVE algorithm. But the SPE+MVE algorithm has two steps :
first, a Semi-definite Programing part and other a SVD (or PCA) part both of which can be approximated. Aim of the
thesis is to analyze these approximation algorithms and to come up with an algorithm to approximate the SPE, from the
point of view of Link Prediction.
PUBLICATIONS <Link>

Adaptive Dictionary Learning for Sentiment Classification & Domain Adaptation accepted at 8th international conference on
Machine Learning and Data Mining, 2012. Authors : R.Mehrotra, R.Agrawal, SA Haider [self-motivated project]
From the abstract : Machine Learning algorithms are often as good as the data they can learn from. Enormous amount of
unlabeled data is readily available and the ability to efficiently use such amount of unlabeled data holds significant
promise in terms of increasing the performance of various learning tasks. In this paper we present a Self-Taught Learning
based framework which uses the K-SVD algorithm for unsupervised learning of sparse representation of data from
unlabeled instances. Using the learnt dictionary, a rich feature representation of the few labeled instances is obtained
which is feeded to a classifier along with class labels to build the model. We evaluate our framework on the task of
sentiment classification and domain adaptation. Both in-domain classification (requiring few very few domain-specific
training instances) and cross-domain classification (requiring 0 labeled instances of target domain and very few labeled
instances of source domain) are performed. The Amazon product review dataset was used for experiment.
PHONE +91 9571 55 2740 EMAIL rushabh0803@gmail.com
ROOM NO. 291, Kr BHAWAN, BITS-PILANI, RAJASTHAN 333 031

Neural Self-Organization based Rectilinear Steiner Minimal Tree Generation in 3D accepted at 14th UKsim conference on
Computer Modeling and Simulation, 2012 Authors :R.Agrawal, R.Mehrotra, A.S. Mandal
Abstract: Given N points in space, generation of Rectilinear Steiner Minimal Tree (RSMT) is always a challenging problem.
As the number of points increases, the complexity of the problem increases exponentially. A self-organization based
method with linear complexity and linear memory requirements has been used for generation of an optimum solution for
a Rectilinear Steiner Minimal Tree in 3D space. An energy equation is formulated to reflect the length of the tree at each
step taking care to ensure that it does not get trapped in a local minimum.

OTHER ACADEMIC PROJECTS <Link>

A compiler for a language (ERPLAG) as part of Programing Languages and Compiler Construction course.
[2012]
Omniscient Bot: Based on the concepts of Cloud Robotics, an Android controlled bot is built which is equipped with
features such as Self Navigation, Speech & Gesture Controls, OCR, face detection
[2011]
Facial Emotion Recogniton : Detects and tracks a human face using the Viola-Jones algorithm, feature extraction done using
Gabor Filters and trained using SVM, subsequently classifying the facial expression as one of the 6 basic emotion types (in
MATLAB).
[2011]
A project on Retail Management System as part of Data Processing course.
[2010]
BlueBuddy [P2P Bluetooth based Social Networking]: An ad-hoc mobile social network is created using P2P communication.
Used the concept of UUID generation and toggling of device between master and slave configuration to enable exchange
of information between every pair of devices.
[2010]
Hand-gesture controlled Robotic Arm: Used Image Processing algorithms to control a 5-degree robotic arm by movement of
fingertips in front of a webcam (without color or any other external markers)
[2010]

ACADEMIC ACHIEVEMENTS AND AWARDS

Gold Medal for Omniscient Bot at APOGEE -2011.


Silver Medal for Swarm Simulation to aid in architectural designing at APOGEE-2011
Silver Medal for Swarm Simulation to aid in architectural designing CyberFiesta-2011
Gold Medal, CyberFiesta (software development competition), APOGEE 2010 for BlueBuddy software.
IIT-Joint Entrance Exam[2008] - A.I.R 5007 (in top 1.5%)
A.I.E.E.E. [2008] A.I.R. of 2108 (in top 0.25%)

POSITIONS OF RESPONSIBILITY
President of the Computer Science Association (2011-2012, 2nd sem) : CSA conducted the first Alumni Research Talks at
BITS Pilani as also its first Hackathon (during APOGEE). Also oversaw the various CSA project and events during
APOGEE-2012.
Project Coordinator at the Computer Science Association (2010-2011) : Involves coming up with various project ideas,
allotting work to team, planning a course of action to ensure completion of work on time, providing technical
guidance/assistance to students and to oversee the progress made in various projects undertaken by students in CSA
throughout the year (focusing mainly on APOGEE and other tech fests)
Joint Secretary of Sangam Cultural Association, a faculty-student body involved in various cultural activities on campus.

PHONE +91 9571 55 2740 EMAIL rushabh0803@gmail.com


ROOM NO. 291, Kr BHAWAN, BITS-PILANI, RAJASTHAN 333 031

Potrebbero piacerti anche