HeXtractor: Automated Heterogeneous Graph Extraction from Multimodal Data
Open-source Python library for transforming structured tabular data and unstructured text into heterogeneous graph representations for Graph Neural Networks.
Here you’ll find a collection of my research and publications in data science, machine learning, and AI.
Open-source Python library for transforming structured tabular data and unstructured text into heterogeneous graph representations for Graph Neural Networks.
Graph neural network architecture for money laundering detection using multiple aggregation mechanisms and custom edge convolutions.
Heterogeneous extension of Graph Isomorphism Networks applied to the FinCEN Files investigation - detection of money laundering patterns.
Application of deep reinforcement learning algorithms to portfolio optimization in complex market conditions, evaluated against traditional optimization strategies.
Evaluation of deep reinforcement learning algorithms for constrained discrete resource allocation in project management using simulation experiments.
Development of a deep hybrid collaborative filtering architecture that integrates content features with collaborative filtering for improved e-commerce recommendation accuracy.