Posts

Community Detection Using Genetic Algorithm and Cuckoo Search

This project was an assignment for an AI course at the Amirkabir University of Technology, where the goal was to identify communities in a network using the Genetic Algorithm and Cuckoo Search. This post will focus on the problem-specific implementations rather than a detailed explanation of the algorithms themselves. First, let’s consider a sample input graph: Finding Communities Using Genetic Algorithm Genetic Representation and Fitness Function: I used a locus-based representation for this problem, as illustrated below: The modularity measure was used as the fitness function. Selection:

Friday, July 29, 2022 | 2 minutes Read

Solving a Real Examination Timetabling Problem

This project was a collaboration with my friend for the combinatorial optimization course, instructed by Dr. Hooshmand. Our task was to create a mid-term exam schedule for the entire MCS department at the AmirKabir University of Technology. The primary challenge was to devise a schedule that accommodated all students, ensuring that no student had more than one exam on any given day and that there were adequate gaps between exams. Given that each student had a unique set of courses, satisfying everyone’s schedule was a complex problem.

Monday, November 15, 2021 | 2 minutes Read

Playing Lunar Lander with Expected Sarsa and a Neural Network

This project is an enhancement of the final project from the Reinforcement Learning specialization on Coursera. It focuses on using the Expected Sarsa algorithm with a neural network as a function approximator to play Lunar Lander. The model was trained using Keras and TensorFlow. The complete source code for this project is available in this GitHub repository.

Sunday, November 14, 2021 | 1 minute Read

Designing an ETL Pipeline for a Data Warehouse

This project was my final submission for a database course instructed by Dr. Gheibi. The project was divided into three main phases: Database Design: The first phase required designing a database for a library, ensuring it adhered to the fifth normal form (5NF). ETL Pipeline: The second phase involved creating an ETL (Extract, Transform, Load) pipeline to synchronize a data warehouse with the operational database. Time Machine: The final phase was to develop a “time machine” feature that could restore the database to a specific point in the past. For the ETL pipeline, a common approach is to replicate each database operation individually, but this can create significant overhead. Instead, I modeled the database as a Directed Acyclic Graph (DAG), where tables and their relationships were represented as vertices and edges. I then used a topological sort of this DAG to determine the optimal order for applying bulk insert, delete, and update operations.

Saturday, November 13, 2021 | 1 minute Read

Detecting Iranian Paper Money to Help People With Visual Disabilities

This project, undertaken just before I began my university studies, was my first application of deep learning to solve a real-world problem. The inspiration came from Microsoft’s Seeing AI app, which offers a range of tools for visually impaired individuals. One of its standout features is the ability to identify currency bills, a functionality I aimed to replicate for Iranian currency, which was not supported by the original app. A significant challenge was the lack of an existing dataset. I dedicated a month to collecting and curating a suitable dataset of Iranian banknotes. During this time, I researched various approaches and decided to use transfer learning with the MobileNet architecture, as the goal was to deploy the model on an Android device. TensorFlow Lite provided the ideal framework for on-device inference.

Friday, November 12, 2021 | 2 minutes Read

Designing a Shooting Simulator Game

Inspired by the annual science fair at my school, which encouraged students to explore new ideas and create innovative projects, my team and I decided to develop a game. Leveraging our experience from the advanced line follower robot project, we designed a shooting simulator game. The setup involved embedding a laser into a toy gun and creating a target box with an archery target. An IP camera was placed inside the box to capture the laser’s position on the target.

Thursday, November 11, 2021 | 1 minute Read

Advanced Line Follower Robot

Building upon our experience with the simple line follower robot, my team and I decided to develop a more advanced version with enhanced capabilities. This iteration replaced the basic IR sensors with a camera, enabling the robot to perform image processing to identify and follow lines and shapes of various colors. Here is a video of the robot: The image processing pipeline was as follows:

Wednesday, November 10, 2021 | 1 minute Read

A Simple Line Follower Robot

This project showcases a simple line follower robot that I designed and built with a team of friends during my junior high school years. The robot’s core is an AVR microcontroller, and its circuit was designed using Proteus. For programming, we utilized Bascom, a straightforward yet capable language that met our project’s requirements. Below is a video demonstration of the robot in action:

Tuesday, November 9, 2021 | 1 minute Read