Fake Review Detection using Machine Learning
Main Page
July 2020 - May 2021
- Implemented a hybrid model for detecting Fake Reviews on eCommerce platforms (e.g., TripAdvisor, Yelp) by combining review-based and reviewer-based approaches for accurate filtering.
- Evaluated the performance of various classifiers, including Naive Bayes, Random Forest, XG Boost, and Support Vector Classifiers, to determine the optimal performance for detecting Fake Reviews.
- Authored a research paper titled “A Review and Reviewer-based approach for Fake Review Detection” in the proceedings of 2021 IEEE 4th International Conference on Electrical, Computer and Communication Technologies (ICECCT), September 2021; summarizing the project's methodology, findings, and experimental results. Contributed to the field of online review analysis. (link )
- Achieved superior performance of 88% compared to existing methods, as demonstrated by robust evaluation metrics
- Technologies used: Python, Natural Language Processing, Machine Learning
Vision-based System for Early Detection of Diabetic Retinopathy
Main Page
July 2019 - December 2019
- Developed an automated model for Diabetic Retinopathy disease detection from eye images, and created a data-efficient model by combining an autoencoder and VGG network in Python.
- Experimented with diverse hyperparameters and optimization functions to optimize the model’s performance.
- Published a peer review paper titled “Reducing Overfitting in Diabetic Retinopathy Detection using Transfer Learning” in 2020 IEEE 5th International Conference on Computing, Communication and Automation (ICCCA), October 2020; presenting the project's methodology, results, and contributions. (link )
- Achieved an accuracy of 80.27% on the selected dataset, surpassing existing methods by 5%
- Technologies used: Python, Deep Learning, Convolutional Neural Networks
Search Engine using Lucene and Hadoop Indexing
Main Page
January 2022 - March 2022
- Engineered a health news blog search engine using a multi-threaded crawler, significantly reducing data extraction time.
- Employed two indexing methods, Lucene in Java using Lucene parser and Hadoop in MongoDB using Map-Reduce functions, achieving comparable indexing speeds with an average processing time of 0.35 seconds per webpage.
- Created a Python Flask-based UI webpage that presents search results with relevant details including title, URL, relevancy score, and a page snippet, enhancing the user experience.
- Technologies used: Python, Java, Lucene, Hadoop, MongoDB
Crime Analysis & Prediction using Big Data
Main Page
September 2022 - December 2022
- Built an ML model for crime prediction in Chicago using historical crime data.
- Leveraged Spark for efficient storage and processing of large-scale data, enabling faster analysis by almost 30%.
- Conducted extensive feature experimentation to identify the most effective predictors, resulting in an improved prediction accuracy surpassing previous methods by 5%.
- Technologies used: Python, Spark, HDFS, Scala, Java
Board Game Robot using Arduino
Main Page
April 2023 - June 2023
- Created a board game robot project that employs AI search techniques, including Minimax, to solve popular board games such as Tic-Tac-Toe, Checkers, and Connect-4 in player-vs-player and player-vs-computer modes.
- Engineered the project using an Arduino microcontroller and appropriate peripherals for input and output, leveraging the power of AI algorithms to make strategic moves and optimize gameplay.
- Demonstrated successful results, achieving an impressive win rate of 95% against human opponents, showcasing the effectiveness of AI in board game scenarios using physical hardware.
- Technologies used: Embedded C/C++, Arduino Programming