Ameed Ahmed

Machine Learning Researcher @Universität Hildesheim

DE

Project 1:Most researches in CV have explored model\'s decision-making process in considerable depth within the general domain.One domain that remains relatively underexplored is historical artworks and portraiture. The aim of this study was to investigate and understand the decision-making process when differentiating between various periodic art style classification (e.g, Baroque, Orientalism) as well as historical portraits. To this end, we primarily employ CS-Win Model as our base model with Evidence Counterfactuals used as XAI, supplemented with findings corroborated by attention maps and segmentation labels extracted via DINOv3.Project 2:Worked on Neural Combinatorial Optimization (NCO) for the Travelling Salesman Problem by improving the training regime for Light Encoder–Heavy Decoder (LEHD) model. Implemented Curriculum Learning & Self-Improvement Algorithm in an attempt to achieve better results. The Self-Improvement Algorithm was aimed at refining heuristically constructed initial solutions without using the ideal labels (generated by complex solvers) for assistance. Achieved an optimality gap of ~0.89%, compared to baseline performance of 1.21%(improvement of 0.32%)

Saint Paul English High School

O levels , Pre-Engineering

Institute of Business Administration

Bachelor of Business Administration - BBA, Marketing + Data Analytics

St. Patrick's High School, Karachi

A Levels, Pre-Engineering

Stiftung Universität Hildesheim

Master of Science - MS, Data Analytics (Focus on ML/AI)