Research Projects in Graduate Level
- Gamma Process Poisson Factorization.
- Active Multitask Learning Using Both Supervised and Latent Shared Topics.
- Noisy Matrix Completion Using Alternating Least Squares Minimization.
- Sparse generative model for analysis of market basket data: The project deals with creating a sparse gen-
erative model for analysis of market basket data where each basket contains only a small subset of the different
items available. The theory draws inspiration from Latent Dirichlet Allocation (LDA), Hierarchical Dirichlet
Process (HDP) and Sparse Topic Model (STM) where Dirichlet distribution, with some specific choice of its
hyperparameters, facilitates sparse modeling in text analysis.
- Improving classification from classifier ensemble and clustering ensemble: } This work aims at building a more consolidated classification from a classifier ensemble and a clustering ensemble. At the core of the mathematical formulation, there is a non-convex function that is optimized using a
new algorithm which, in principle, is similar to block co-ordinate descent type algorithm.
- Customer segmentation, churn modeling and predictive modeling of smartphone customers:
For a South Korea based telecommunication company.
- A survey on Consensus Clustering: This survey explores all the existing methodologies for combining multi-
ple clusterings and groups them according to the approaches adopted. A possible extension of this work is to
consider cluster ensemble under pairwise constraints.
Research Projects in Undergraduate Level
- Improving the performance of existing Ant System algorithms like basic Ant System, Elitist Ant System, Max-min Ant System and Rank-Based Ant System, Designing of fractional order PID controllers using stochastic optimization algorithms like Particle Swarm Optimization and Differential Evolution and Recognition of regular shaped geometric contours from images using a newly proposed ant system algorithm under Dr. Amit Konar, Professor, Department of Electronics and Telecommunication Engineering, Jadavpur University, Kolkata, India.
- Final Year project under Dr. Amit Konar on Motion planning and mapping of a surface using Khepera Minirobot. The purpose of this research project is to accurately map an unknown environment using the Khepera minirobot. Sensor readings are obtained for various positions and orientations of the robot. The world map is formed using fuzzy logic. The robot is required to navigate by the optimal path, using discrete stochastic optimization algorithms.
- Summer Project 2008 on Delaunay Tessellation based method of faster matching spatial point set pattern under Dr. Bhargab B. Bhattacharya, Professor, Advanced Computing and Micro-electronics Unit, Indian Statistical Institute, Kolkata, India. This work has many real life applications like recognition of _nger prints based on minutiae location, image matching under rigid motion and distortion etc. Aspects of combinatorial geometry like Voronoi Diagrams and Delaunay Triangulation have been used in modeling the algorithm.
- Summer Project 2007 on Development of automatic dimension acquisition kit under Dr. Sudipta Mukhopadhyay, Assistant Professor, Department of Electrical and Electronics Engineering, Indian Institute of Technology, Kharagpur, India. This project was highly successful and the work was published later in 2nd International Conference on Cognition and Recognition (ICCR 2008). The real life applications of the project include measuring the dimensions of the luggage placed on the conveyor belt in an International Airport and object assembly using robots.
- Energy optimization in wireless sensor networks using Ant Colony Optimization algorithm under Dr. M.K.Naskar, Professor, Department of Electronics and Telecommunication Engineering, Jadavpur University, Kolkata, India. The work was published in one International conference and one International Journal.