Machine Learning
Network Security
Bioinformatics
Robotics
H. Abid, N. J. Jenny, S. M. Shovan, “Improved Identification Performance of Lysine Glycation PTM using PSI-BLAST”, IEEE Tensymp 2020, Dhaka, Bangladesh, June 2020. [Full Paper] [Free Version]
Enhancing Real-Time Application Performance by Predicting Congestion in Data Transport Networks Using Machine Learning (Master's Thesis) [2024]
The massive increase in 5G usage for video streaming can cause short but intense spikes in data transfer rates, potentially causing network congestion. Congestion detection and prediction are essential for ensuring user satisfaction and enhancing the performance of real-time applications. By leveraging historical network measurement data such as Round Trip Time (RTT), Jitter, and packet loss ratio, machine learning algorithms can analyze traffic patterns, predict congestion based on the attributes, and provide adaptive solutions. We have focused primarily on ML models due to their adaptability to learn from historical patterns, a capability where traditional rule-based methods may fall short. Furthermore, past researchers faced difficulties in obtaining measurement data for fully comprehensive real-world networks. We have collected real-world network measurement datasets over a six-month period for a complete network. Using the datasets and topology information, we trained Support Vector Regression (SVR), Long Short-TermMemory (LSTM), and XGBRegressor (Extreme Gradient Boosting) to forecast the RTT values of each router. To predict network anomalies based on packet loss and high RTT values, Random Forest and Support Vector Classifier (SVC) models have been developed. By analyzing the outputs of these two models, a method is proposed to predict network congestion. While this study is limited by data quantity and the use of simpler ML models, it offers a comprehensive solution and a solid foundation for predicting network congestion symptoms. Network congestion prediction can be further improved by exploring areas such as model ensembling, explainability in deep learning models, transfer learning approaches, and integrating multi-source data.
Project Link Project Link 2 Project Report
Smart Sailor [2022-2023]
The goal of the Smart Sailor project is to develop a situational awareness platform and collision avoidance algorithms for Åboat which is an autonomous and remote-controlled research boat platform being developed by the IT department at Åbo Akademi. Currently, we are using the simulated version of Åboat to build our system in AILiveSim. To develop situational awareness, we are combining lidar data with camera images to determine the precise distance of nearby objects. By implementing reinforcement learning on the AiLiveSim data, we are planning to build the collision avoidance algorithm.
Performance Measures of Different Solution Approaches for the Dynamic and Stochastic Vehicle Routing Problem (Bachelor's Thesis) [2016-2017]
Customers in freight transport desire more flexibility and fast fulfillment of their orders. The advances in communication technology permit to store and analyze huge amounts of data and also help to serve customers in real-time. This motivates a new version of the vehicle routing problem, the so-called Dynamic and Stochastic Vehicle Routing Problem (DSVRP). There are many approaches which solve DSVRP. Among them Multi-Scenario Approach (MSA) and Stochastic Programming with Recourse (SPR) are well known. The goal of this project is to compare between these two approaches on basis of travel cost and computational time. To achieve the goal, generating benchmark data sets for DSVRP and solving the approaches are vital part of this project.
Supervisor: Prof. Dr. Jürgen Pannek
Sentiment Analysis of imdb movie reviews using CNN, LSTM hybrid model [March 2019 - April 2019]
Nowadays, Social media has become a great source of getting user's opinion. By using imdb movie reviews as our benchmark data set, we wanted to predict whether a opinion is good or bad. We used Word Embedding for feature extraction. To train the model, first we used CNN and after that LSTM which improved our performance than using only LSTM. We trained our model with different vocabulary size to compare the performance.
Supervisor: Sajid Ahmed
Prediction of Lysine Glycation PTM Site in Protein Sequence using PSI-Blast Feature Extraction [July 2019 - November 2019]
Protein being an important component in human body, undergoes enzymatic modifications referring as Post-transational modification (PTM) which forms a mature product of protein. Glycation is one of the 40 ptms that is discovered so far, a non-anzymatic covalent bonding of sugar and protein or lipid. It is a biomaker for renal failure, diabetes and implicated in other health issues as well, imprinting a significant importance for its site identification in a protein sequence. In our experiment, we intended to improve Glycation prediction using a new feature extraction technique, called PSI-BLAST. We used SVM and Random forests classifier to train the model.
Supervisor: S.M. Shovan