Pulkit Agrawal

E-mail: abqchp@r.postjobfree.com Phone: +1-412-***-**** Home Page: http://cs.berkeley.edu/~pulkitag

EDUCATION

University of California, Berkeley (Aug 2011 to present)

Ph.D. Student in Computer Science

GPA: 3.92/4.0

Indian Institute of Technology, Kanpur (2007-2011)

Bachelor of Technology in Electrical Engineering

GPA: 9.4/10.0

.

ACADEMIC AWARDS

International Fulbright Science and Technology Award

o

Director s Gold Medal for best all round achievement and leadership in graduating class of 2011 at IIT Kanpur.

o

Academic Excellence Award for the year 2007-08, 2008-09 and 2009-10 IIT Kanpur, for distinctive academic achievements.

o

rd

Sridhar Memorial Prize for best student in Electrical engineering department based on academic performance at the end of 3 year.

o

rd

Smt. Saraswati Singh Scholarship for best student in Electrical engineering based on GPA at the end of 3 year.

o

Goldman Sachs Global Leadership Award 2009

o

OP Jindal Engineering and Management Scholar (OPJEMS) 2009

o

First Prize in paper presentation competition, Eureka in Techkriti- 10, National Level Intercollegiate technical festival.

o

First Prize in Electromarket, Digital & Analog circuit design competition in Techkriti-10.

o

Runner-up, in Prayog, Experimental Science Competition, Techkriti 09.

o

Runner-up in Advanced level, embedded circuit design competition Techkriti-08.

o

RESEARCH INTERESTS

Artificial Intelligence including Machine Learning, Computer Vision, Robotics, Computational Neuroscience, Neural Networks and

understanding how does the brain work.

PUBLICATIONS

The automatic assessment of knowledge interaction processes in project teams, G. Gweon, Pulkit Agrawal, M. Udani, B. Raj, C. Rose,

International Conference of Computer Supported Collaborative Learning (CSCL) 2011 (Best Student Paper Award)

.

EXPERIENCE

Research Intern, School of Computer Science, Carnegie Mellon University, USA (May to July 2010)

Automatic assessment of student reasoning process in face to face interaction using speech Mentor: Prof. C. Rose & Bhiksha Raj

Given the speech recordings of N speakers in a conversation, the goal was to automatically detect statement s when speakers were

o

saying something important (i.e. involving reasoning) using acoustic features only (without speech to text conversion).

Given the skewed nature of the data, remarkably, precision of 0.51 and F-Score of 0.56 was achieved.

o

Gaver s Hypothesis testing Mentor: Prof. Bhiksha Raj

Two models, a Multi Gaussian mixture model and another employing a daboost were developed for computationally validating the

o

Gaver s Hypothesis for environmental sound classification.

Student Intern, University of Melbourne, Australia (May to July 2009)

Estimation of Rosacea using Image Processing Mentor: Prof. Jonathan Manton, Dep of Elec. Eng.

A colour based skin model capable of objectively tracking changes in severity of skin disease Rosacea from photographs of pat ients

o

taken over a period of few months while they were under medication was developed and tested.

KEY PROJECTS

A Probe into decoding brain activity using fMRI and MEG Mentor:Dr. Jack Gallant, Dep. of Psychology, UC Berkeley

Implemented ridge regression models to predict voxel responses for given visual stimuli with high correlations.

o

Explored combining information between MEG and fMRI using canonical correlation analysis, regression techniques etc.

o

Collocation based approach for training Recurrent Neural Networks Mentor: Dr. P. Abbeel, EECS, UC Berkeley

Formulated training of RNN as a trajectory optimization problem. Collocation method was implemented using cvx and mosek. The

o

training turned out to be slower than back-prop for simple networks and failed for more complex ones.

Scene Analysis using Trajectory Clustering in Surveillance Videos Mentor: Dr. A. Mukerjee, Prof., Dep. of Comp. Sc. Eng., IIT Kanpur

Proposed a cognitive experiment to measure relative effects of speed, orientation and locatio n on similarity of trajectories.

o

Implemented hierarchical clustering incorporating the same metric to obtain usual and unusual trajectory clusters.

o

Scene Classification Mentor:Dr. Simant Dube, Dep. of Comp. Sc., IIT Kanpur

Designed a classification scheme in form of a decision tree based on linear discriminant analysis for identifying 8 image classes.

o

89% accuracy for Artificial v/s Natural categorization, 71% for classification of highways and 60.2% overall accuracy was obtained.

o

Modeling Responses of V1 Neurons to Natural Vision Movies Mentor: Dr. A. Mukerjee, Prof., Dep. of Comp. Sc. Eng., IIT Kanpur

Method of reverse correlation was used for estimating spatio temporal receptive fields (STRF) of individual neurons in response to

o

the stimuli of natural vision movies. Average correlation of 0.16 was obtained between actual and predicted spiking activity.

OTHER PROJECTS

Implementation of back propagation, BPTT, RTRL in design Artificial Neural Networks.

o

Robust Two Degree of Freedom Vehicle Steering Controller Design.

o

Autonomous Line following Robot.

o

Micro-controller based Chat Client capable of communicating with desktops.

o

RELEVANT COURSES AND SKILLS

Statistical Learning Theory, Advanced Robotics, Computer Vision*, Convex Optimization*, Scalable Machine Learning*, Digital

o

Signal Processing, Visual Neurophysiology, Cognitive Sciences, Probability and Statistics, Artificial Neural Network, Image

Processing, Applied Stochastic Processes, Data Structures and Algorithms.

Programming and toolboxes: C++, Matlab, Python, Java, cvx, Mosek, Open CV

o

(*Will be completed at end of Spring 2012)

LEADERSHIP

General Secretary, Science & Technology Council, IIT Kanpur (Elected) (2010-11)

o

Vice-Captain, Institute Aquatics Team (IIT Kanpur) (2009-10)

o

Coordinator, Cryptography Contest-Techkriti-09 (Inter-Collegiate Technical Festival of IIT Kanpur) (2009)

o

Secretary, Ritambhara, The Fashion Show Antaragini (Inter-Collegiate Cultural Festival of IIT Kanpur) (2008)

o

Student Guide, Counselling Service, IIT Kanpur (2008-09)

o

REFERENCES

Dr. Jack Gallant (abqchp@r.postjobfree.com) Dr. Bhiksha Raj (abqchp@r.postjobfree.com)

Department of Psychology School of Computer Science

University of California, Berkeley Carnegie Mellon University

Dr. Jonathan Manton (abqchp@r.postjobfree.com)

Future Generation Professor

Department of Electrical and Electronic Engineering

University of Melbourne, Australia

Contact this candidate