Fardin Abbasi's Projects
Foundation is a flexible, modular, and composable framework to model socio-economic behaviors and dynamics with both agents and governments. This framework can be used in conjunction with reinforcement learning to learn optimal economic policies, as done by the AI Economist (https://www.einstein.ai/the-ai-economist).
Implementation and analysis of a Ball & Beam Control System, including system modeling, linearization, controllability, observability, and the design of controllers and observers, with a focus on stabilizing the system using PID controllers, state feedback, and estimators to handle nonlinearities, disturbances.
Performing classification of penguin species using a Naive Bayes classifier, both implemented from scratch and with Scikit-Learn, including data preprocessing, training, and evaluation.
Implementation of Decision Tree and Random Forest algorithms, with various hyperparameters, developed from scratch and using scikit-learn for comparison and analysis.
Exploring deep reinforcement learning techniques, particularly focusing on the implementation and testing of algorithms like Deep Q-Learning (DQN) and Dueling DQN.
Modeling and analyzing a magnetic levitation system, including the design and evaluation of PID and PI controllers to achieve stability and desired performance in both time and frequency domains.
Implementing the Expectation-Maximization algorithm and applies Gaussian Mixture Models (GMM) to classify images.
About Me
Implementation of a 2D correlation method for locating an electrical element (BA3240) on a PCB.
Implementations and analyses of various mathematical and computational techniques, including Lagrange interpolation, LU decomposition, image compression with SVD and FFT, image denoising, histogram matching, and QR decomposition using Gram-Schmidt methods.
Modeling and simulating electrical and mechanical systems using the Laplace transform in MATLAB and Simulink.
Operational Research [ECE 145] - Fall 2023- University of Tehran - Dr. Ramezani Moghaddam
Performing polynomial regression of varying degrees on data affected by white and Poisson noise, evaluating the model performance based on MSE loss and the bias-variance trade-off.
Implementing logistic regression with L2 regularization from scratch to classify circular datasets by mapping the feature space into higher dimensions.
Containing various ML projects, including image and text classification, clustering, regression, and neural networks. Projects include implementations of Random Forest, SVM, Decision Trees, EM algorithms, and advanced models like MLP and CNN, with datasets ranging from facial emotion detection to ECG signals and more.
Detecting fake images using machine learning, involving feature extraction from real and AI-generated images, and implementing various classification models such as Random Forest, SVM, and Logistic Regression to accurately distinguish between real and fake images.
Machine Learning Course [ECE 501] - Spring 2023 - University of Tehran - Dr. A. Dehaqani, Dr. Tavassolipour
Two C++ projects demonstrating Object-Oriented Programming (OOP) principles.
Performing non-parametric Parzen Window density estimation with a Gaussian kernel, implementing the method both from scratch and using built-in functions, and compares the results across various window widths and sample sizes.
Applying Principal Component Analysis for image compression, exploring how varying numbers of principal components affect image quality and compression ratio.
Advanced Programming [ECE 119] - Spring 2024- University of Tehran - Dr. Khosravi
Engineering Probability and Statistics [ECE 092]- Fall 2021- University of Tehran - Dr. A. Dehaqani
Analyzing random processes to determine stationarity using statistical methods and implements analog-to-digital conversion, quantization, and signal transmission with noise in MATLAB.
Advanced Programming [ECE 119] - Spring 2024- University of Tehran - Dr. Khosravi
Digital Systems 2 Course [ECE 778] - CA3 - Spring 2023 - University of Tehran - Dr. Safari
Design and implementation of RISC-V processor with a pipelined datapath, controller, and hazard unit.
Design and implementation of RISC-V processor with a single-cycle datapath and controller.
Implementing a convolution function from scratch and analyzing an LTI system by adding noise and echo to an audio signal, with detailed plots and results provided.
Implementations of basic components like timers, LCD displays, external interrupts, and digital pins on the STM32F103R6 microcontroller, demonstrating fundamental embedded system concepts.
Implementation of Support Vector Machines on the Iris dataset, featuring both scratch and scikit-learn versions with various kernels.