Shubham Goel IIT Bombay

Email : [email protected] Computer Science & Engineering

Github : https://github.com/shubham-goel 4th Year Undergraduate
Phone : +91-8879069817 GPA : 9.63/10.0

Interests
Computer Vision, Artificial Intelligence, Logic for Computer Science

Publications
[1] Pushing The Envelope for Boolean Function Synthesis+
S. Akshay, Supratik Chakraborty, Shubham Goel, Sumith Kulal, Shetal Shah
CSE IITB Technical Report (TR-CSE-2017-87)
[2] Computing Scores of Forwarding Schemes in Switched Networks with Probabilistic Faults
Guy Avni, Shubham Goel, Thomas A. Henzinger, Guillermo Rodriguez-Navas
23rd International Conference on Tools & Algorithms for the Construction and Analysis of Systems (Tacas),2017
+
Names of authors sorted alphabetically by last name. Abridged version, full preprint available on request.

Scholastic Achievements
· Secured All India Rank 6 in JEE Advanced 2014 among 150 thousand candidates
· Secured All India Rank 50 in JEE Mains 2014 among over 1.3 million candidates
· Awarded the AP Grade for exceptional performance in courses Logic for CS, Digital Logic Design, Engineering
Drawing and Biology
· Received the Institute Academic Award, IIT Bombay for exceptional academic performance in 2014-15

Olympiads & Scholarships

· Silver Medalist at the 46th International Chemistry Olympiad, Hanoi, Vietnam
· Best Theorist and Experimentalist at the Orientation cum Selection Camp for the 46th IChO
· Received the Best Solution Award at the National Selection Camp for the 54th IMO, Colombia
· Recipient of the KVPY (Kishore Vaigyanik Protsahan Yojna Fellowship) in 2013 by Govt. of India
· Awarded the NTSE (National Talent Search Examination) Scholarship in 2010 by N.C.E.R.T. New Delhi

Research Experience
Estimating Dense Correspondences on Wide Baseline Images Ongoing since Autumn 2017
Guide: Prof. Arjun Jain, IIT Bombay
· Using data driven approaches for finding dense correspondences between images with scale and viewpoint changes
· Have taken a 2-step hierarchical approach by first predicting a coarse match, and then a fine one
· Use robust descriptors and a correlation volume encoding descriptor similarity between points and coarse regions
in the first and second image respectively
· Preliminary experiments gave promising results, beating state-of-the-art tools like DeepMatching by 10%
· Currently working on reducing memory intensiveness and improving network architecture to improve performance

BFSS: Boolean Functional Synthesis Ongoing since Spring 2017

Guide: Prof. Supratik Chakraborty, IIT Bombay
· Working on a CEGAR+AIG based state-of-the-art tool (BFSS) for the Boolean Functional Synthesis problem
· Devised an algorithm reducing the total number of refinements required at the cost of a few extra SAT calls
· Built a synchronous system wherein the solver runs in the background and pipelines counterexamples to BFSS;
Cut-shorts on waiting time and provides probabilistic guarantees on correctness of the Skolem Functions
· Worked on writing a full length research paper which will be under submission soon in CAV 2018
· Currently working on improving the algorithm even further by exploiting AIG structure
Restoration of Manifold-Valued Images Summer 2017
Guide: Prof. Stefan Roth, TU Darmstadt
· Worked on restoration (denoising and inpainting) of images that take values in Riemannian manifolds
· Explored the use of higher order differences and arbitrary filters in modeling the loss objective.
· Came up with a family of frameworks for applying arbitrary zero-sum filters to manifold valued patches; Provided
flexibility and generalized previously defined first and second order differences over manifolds
· Implemented this framework, reproduced previous work and built an image denoiser in Matlab that works with
multiple arbitrary filters

Forwarding Schemes in Switched Networks with Probabilistic Faults Summer 2016

Guide: Dr. Thomas Henzinger, IST Austria
· Assigned scores to forwarding schemes indicative of their robustness towards link crashes and faults
· Designed and implemented a reduction of the scoring problem to SAT counting
· Implemented different iterative and statistical approaches for scoring propositional rule based forwarding schemes
· Proved #P-completeness of the scoring problem, started work on complexity of approximate scoring algorithms
· Gave a talk at IST Austria regarding the same

Nodal Domains of Eigenfunctions of Quantum Billiards Summer 2015

Guide: Dr Sudhir Jain, BARC
· Made analytical attempts for solving the Helmholtz equation for the 60°-120°rhombus
· Numerically solved for eigenfunctions of the 60°-120°rhombus using the Method of Fundamental Solutions (MFS)
· Developed a modification to the Hoshen-Kopelman Algorithm for counting nodal domains

Academic Projects
Real-Time Tracking of Non-Rigid Objects — Computer Vision
· Implemented the mean shift algorithm for tracking non-rigid objects in a video sequence
· Extended the algorithm to handle huge scale changes
· Implementation was robust to blur, deformation and partial occlusions

Vector-Valued Image Regularization with PDEs — Fundamentals of Digital Image Processing

· Built an image regularization tool using techniques based on solutions to Oriented Laplacian PDEs
· Used a generic anisotropic diffusion equation which provides a simple interpretation of the regularization process
in terms of local filtering with spatially adaptive Gaussian kernels
· Applied the solution for colour image smoothing, colour image inpainting, and flow visualization

tusSAT: A FPGA based SAT solver — Digital Logic Design

· Designed a VHDL package for representation of atomic variables, clauses and expressions
· Implemented a modification of the DPLL algorithm alongside heuristics for variables selection
· Testing suite built from DIMACS Implementation Challenge: Satisfiability
· Featured among other SAT Solvers on satlive.org

Movie Recommendation Engine — Foundations of Machine Learning

· Developed a movie recommendation engine in Python using popular collaborative filtering techniques
· Primarily based on the research done on Single Value Decomposition method during the Netflix Prize competition
· Implemented and tested other machine learning techniques like Baseline predictor and k-Nearest Neighbor Model

Technical Skills
Programming Languages Fluent in C++, Lua, Python; Familiar with Java, VHDL
Libraries Torch, NumPy, TensorFlow, Scikit-Learn
Software Skills git, MATLAB, GNU Octave, AutoCAD, LATEX, CMake
Web Development HTML, CSS, JavaScript, PHP, Laravel (PHP), Django, MySQL, PostgreSQL
Positions of Responsibility
Teaching Assistant, IIT Bombay
· MA 105 : Advanced Calculus under Prof. I.K.Rana Autumn ’15
· CS 226 : Digital Logic Design under Prof. Supratik Chakraborty (Awarded TA of the Month) Spring ’17

Mentor, Institute Student Mentorship Programme 2017 - Ongoing

· Responsible for guiding 12 freshman, focussing on academics and holistic development
· Providing counsel and helping them adjust to campus life

Mentor, Department Academic Mentorship Programme 2017 - Ongoing

· Mentor to 6 students for their academic and general concerns, and helping them cope with the curriculum
· Mentor to additional 2 students in academic rehabilitation program (ARP), and helping them get back on track

Academic Resource Person, 46th IPhO, Mumbai 2015

· Responsible for grading the theory papers of participants from 87 Countries
· Responsible for moderation of marks with Leaders from participating countries

Web Convener, Student Technical Activity Body, IIT Bombay 2015-16

· Developed portals for documentation and registration of participants of STAB events
· Responsible for maintaining the STAB website, modifying content and improving functionality

Extracurriculars
· Ranked 14th in ACM ICPC Chennai Onsite contest and 20th in the Online Regionals in 2016
· Winner in Microsoft’s code.fun.do Finalist Forum from amongst 53 teams across 15 colleges
· Qualified for the onsite finals of Microsoft’s Build The Shield, a network security competition
· Won Bronze medal in Table Tennis General Championship, IIT Bombay in 2015
· Secured 4th position in an autonomous line-following bot making competition organized by Electronics Club, IITB
· Successfully completed a 1 year of social service under the National Service Scheme IIT Bombay
· Attended the Vijyoshi National Science Camp in 2013 organized by Indian Institute of Science, Bangalore
· Represented District Hisar at the State Level Championship in Inline Roller Skating in 2007,2008