Resume
TL;DR – This isn’t official one.
Master’s Dissertation (Thesis)
Comparative Analysis of Supervised Learning Methods for Fake News Detection
Supervisor : Dr. Alan Godfrey
Abstract :
Fake news detection is a challenging problem, and there are many different artificial intelligence approaches developed to detect fake news. However, it is still unclear which approach is better and where further research should lead. To accomplish this goal, we conducted this study to compare the existing supervised learning methods for classifying fake news. We conducted our study on ‘Liar, Liar, Pants on Fire’ dataset with 12.8K columns. We implemented NLP technique TF-IDF and evaluated the performance of seven machine learning algorithms i.e., Decision Tree, Random Forest, Naïve Bayes, XGBoost, Logistic Regression, Support Vector Machine (SVM) and LinearSVC and six deep learning methods namely Artificial Neural Network (ANN), Convolutional Neural Network (CNN), Recurrent Neural Network, Long Short-Term Memory (LSTM), Gated Recurrent Units (GRU) and Bi-Directional LSTM. Our results reveal that machine learning algorithms like Random Forest, Logistic Regression and SVM and deep learning methods like LSTM and GRU are the most efficient for fake news detection task, whereas Decision Tree, LinearSVC, Artificial Neural Network, Recurrent Neural Network are least efficient. We have also presented evaluation of our implementation and model performance. Finally, we discussed how the fake news detection problem can be approached in future for better results.
$→$ Presentation on MSc Dissertation given at University of Southern Denmark for PhD in Reinforcement Learning with Neural Stochastic Processes : Link
$\rightarrow$ Fun Fact : Once scored All India Rank 721 in National Science Talent Search Examination (NSTSE)
Additional Learning (Coursera)
| Specialization | Offering University | Duration | Certificate |
|---|---|---|---|
| Machine Learning Operations (MLOps) Specialization | Duke University | 112 Hours | Link |
| Applied Kalman Filtering Specialization | University of Colorado System | 82 Hours | Link |
| Explainable AI (XAI) Specialization | Duke University | 33 Hours | Link |
| Mathworks Computer Vision Engineer Professional Certificate | Mathworks | 98 Hours | Link |
| Tensorflow 2 for Deep Learning Specialization | Imperial College London | 105 Hours | Link |
| Advanced Statistics for Data Science Specialization | John Hopkins University | 40 Hours | Link |
| Reinforcement Learning Specialization | University of Alberta | 80 Hours | Link |
| Writing in the Sciences | Stanford University | 30 Hours | Link |
| IBM AI Engineer Professional Certificate | IBM | 118 Hours | Link |
| Google Data Analytics Professional Certificate | 240 Hours | Link | |
| Programming in C++ Specialization | Codio | 40 Hours | Link |
| DevOps on AWS Specialization | AWS | 36 Hours | Link |
| Data Visualization & Communication with Tableau | Duke University | 24 Hours | Link |