Scholar
Melika Ayoughi
Google Scholar ID: hyyVyt4AAAAJ
PhD University of Amsterdam
Citations & Impact
Citations
8
H-index
2
i10-index
0
Publications
7
Co-authors
12
list available
Publications
Resume (English only)
Academic Achievements
- - Paper 'Designing Hierarchies for Optimal Hyperbolic Embedding' published at ESWC2025 conference and nominated for Best Research Paper Award
- - 'Continual Hyperbolic Learning of Instances and Classes' under review, proposing a novel hyperbolic continual learning algorithm
- - 'PART: Self-supervised Pretraining with Pairwise Relative Translations' also under review
Research Experience
- - Completed a research internship at Apple Machine Learning Research, collaborating with Hanlin Goh on self-supervised pretraining of transformers using relative patch transformations
- - During her master's, interned at Dexter Energy Services for weather nowcasting using satellite imagery
Education
- - PhD: University of Amsterdam, supervised by Prof. Dr. Paul Groth (INDE Lab) and Dr. Pascal Mettes (VIS Lab), focusing on Hierarchical Knowledge in Multimodal Representation Learning
- - Master's: University of Amsterdam, Artificial Intelligence, master’s thesis collaborated with TomTom, focused on object detection under high class imbalance
- - Bachelor's: University of Tehran, Computer Science with a focus on Information Technology, interned at ETH Zurich's Computational Social Science Lab, working on large-scale distributed systems (DIAS & EPOS), emphasizing data management
Background
- - Research Interests: Computer vision, specifically multimodal data such as video, audio, text, and graphs
- - Professional Field: Hierarchical Knowledge in Multimodal Representation Learning
- - Brief Introduction: Aiming to improve high-level understanding of videos, such as object-centric learning and relationship detection, and ease retrieval by storing information in a knowledge graph.
