Danial Saber
Graduade Research/Teaching Assistant
About Me
Hello! I’m Danial Saber. I am passionate about researching on Deep Learning with a particular focus on Graph Neural Networks. My main programming language is Python and I am proficient in popular frameworks such as Pytorch and Pytorch Geometric. I am also interested in applied machine learning, natural language processing, and AI in healthcare. I am a quick learner and a team worker that gets the job done.
Currently, I am a graduate research/teaching assistant at the SAIN Lab under the supervision of Dr. Amirali Abari at Ontario Tech University.
Bio
Professional Skills
Work Experience
Big Data Analytics: Conducted tutorials, led programming labs (Python), and graded assignments. Instructor: Dr. En-Shiun Annie Lee.
Education
Selected Courses: Survey in Computer Science Research Topics & Methods, Machine Learning, Affective Data Science, Collaborative Design and Research, IT Security
GPA: 4.18 / 4.30
Projects
Python / Pytorch / PyG
SE2P: Scalable Expressiveness through Preprocessed Graph Perturbations
Proposed SE2P, a model combining flexibility, scalability, and expressiveness. The approach offers four configuration classes, each offering a unique balance between scalability and generalizability. Extensive experiments were done for graph classification tasks using PyTorch and PyTorch Geometric. Accepted in CIKM 2024.
Python / Pytorch / PyG
Fake News Detection on Twitter
Implemented various Graph Neural Network models, including Graph Convolutional Network (GCN), Graph Attention Network (GAT), Graph Isomorphism Network (GIN), and GraphSAGE, to evaluate and compare their performance against Multi-Layer Perceptron models in a graph classification task, where Twitter news and corresponding retweets are represented as graphs.
Python / Sklearn / Matlab
Developing a global approach for determining the molar heat capacity of deep eutectic solvents
Proposed a universal model based on least-squares support vector regression with the Gaussian kernel function (LSSVR-G) to estimate the molar heat capacity of deep eutectic solvents. This LSSVR-G has been chosen among fourteen intelligent and two regression-based methodologies.