Projects

Reddit Conflict Mediation Recognition

Under the guidance of Prof Susan Fussell, Cornell University

• Implemented an algorithm to early detect conflicts in the Reddit community using different kernelized Support Vector Machines model over k-means based Reddit user types clusters using Python and C++.

Bilingual Politeness Perception Recognition

Under the guidance of Prof Cristian Danescu, Cornell University

• Built a supervised learning pipeline to disambiguate politeness perceptions for Mandarin and English speakers
• Implemented a specialized probabilistic clustering algorithm to identify features for politeness detection.

Machine Translation using Recurrent Neural Networks

• Implemented an RNN model from scratch with encoder and decoder models that incorporated attention mechanisms for solving sequence to sequence neural machine translation tasks.

Sentiment Analysis of multilingual code mixed data using Deep Learning Algorithms

(Under the guidance of Prof. Yashvardhan Sharma (Associate Prof., Dept of Computer Science, BITS Pilani))

• Worked under Dr. Yashvardhan Sharma (Prof. CSIS BITS Pilani), for developing Machine Learning and Deep Learning models for Sentiment Analysis of code mixed data for different Indian languages.
• Deployed Convolutional Neural Networks (CNN) and Long Short Term Memory (LSTM) and Recursive Neural Networks (RNN) models for the purpose of sentiment analysis.
• Generated and published SentiWordNet dictionaries for 13 different Indian languages, which have the information of both the polarity scores along with the Part Of Speech (POS) tag.
• Implemented different machine learning algorithms for language identification and used the SentiWordNet dictionaries for computing sentiments of Code-Mixed data of Indian languages.
• Participated in the shared task of Question Classification of Code-Mixed data (English+Bengali), in Forum for Information Retrieval (FIRE) conference, and secured 2nd highest accuracy using different machine learning algorithms.
• Most of the work involved python and C++ scripting, using Theano and Tensorflow framework for deep learning models deployment.

Modification of AODV protocol using FireFly Optimisation Algorithm

(Under the guidance of Prof. Rahul Banerjee (Prof., Dept of Computer Science, BITS Pilani))

• Worked on optimisation of Adhoc On-demand Distance Vector Routing protocol in a simulated Vehicular Adhoc Network (VANET).
• Used bio-inspired FireFly algorithm with weighted delays and hop count for optimising AODV protocol.
• Incorporated link-lifetime as a measure for detecting route-reliablity in simulated VANETs.
• Developed highway models using Network Simulator-3 as a tool and simulated the optimised AODV protocol this, using C++ as a programming language.

Data Provenance modelling using Graph Databases and Role in Big-Data Analytics

(Under the guidance of Prof. Navneet Goyal (H.O.D., Dept of Computer Science, BITS Pilani))

• Worked on analysing the role of provenance in data analytics and implemented methods for converting a relational (Oracle) to graph database (Neo4j).
• Built modules for capturing provenance including Data Provenance for Historical Queries Framework (DPHQ) which captures and stores the processed queries and Query and Provenance Capturing Module (QPCM).
• Developed a Zero Information Loss Database over the relational db), and modeled the stored provenance in form of a directed acyclic graph.
• Used Cypher for querying the graph database (Neo4j) and java for writing the modules.

Compiler Construction

(As a part of Compiler Design course under Prof. Vandana Agarwal (Assistant Professor, Dept. of Computer Science, BITS Pilani))

• Developed LL(1) Compiler in C, for a strongly typed language, supporting arithmetic and logical operations over primitive and record type variables.
• Prepared different modules including:
  • Grammar Design for LL(1) language
  • Lexer, Parser and Error checking and recovery modules.
  • Semantic Analyser with the use of Symbol Tables and Abstract Syntax Trees (AST).
  • MASM compatible assembly code generation.
• The compiler was developed using C as a programming language.

MAC Layer modelling and modifications in IEEE 802.11 Beacon Frame

(Under the guidance of Prof. Vishal Gupta (Associate Prof., Dept of Computer Science, BITS Pilani))

• Developed models for MAC layer implementations at hardware and software level, in order to modify fields of Beacon Frame and make them accessible to the application layer.
• Developed an Overlay MAC Layer over 802.11 MAC layer using loosely synchronized clocks and distributed algorithms for time slot allocation.
• Used the approach of Remote MAC technique as implemented in the CAPWAP protocol.
• Modified the android kernel so as to access thefields of IEEE 802.11 BeaconFrame from an android based device.

Predictive Modelling for Multivariate Time Series Data

(As a part of Machine Learning under Prof. Navneet Goyal (H.O.D., Dept of Computer Science, BITS Pilani))

• Developed a model for prediction of signs using Australian Sign Languages (AUSLAN) dataset as a part of Machine Learning course.
• Used Singular Value Decomposition for dimensionality reduction and Dynamic Time Warping Algorithm for classifying sign with an accuracy of 96.6%

Linux Kernel Modification

(Under the guidance of Prof. Vishal Gupta (Associate Prof., Dept of Computer Science, BITS Pilani))

• Worked on introducing code in the kernel by the technique of using Dynamic Kernel Linkers by means of provoking system calls to implement desired functionalities.
• Modified Device Drivers and implemented them on Linux based platforms, which covered specifications about Network Drivers.

Security System for password based door locking

(As a part of Microprocessors and Interfacing course under G Sai Sesha Chalapathi(Asst Prof, Dept. of Electrical & Electronics Engineering, BITS Pilani)

• Developed, designed and implemented a security system for password enabled door locking with burglar alarm using 8051 Micro-controller and 8086 Assembly language programming.

Handwritten Character Recognition & Android app Queriea beta version

(As professional assistant, SDET Labs, BITS Pilani)

• Developed character recognition system using Machine Learning techniques (Decision Trees) for the e-complaint android app Queriea.