← Return to projects

project

Detecting AI-Generated Code

Comparing three method families ranging from traditional classifiers, to graph-based methods, to deep learning embedding-based classification.

01
research
02
Research Lead
03
Python / PyTorch / PyTorch-Geometric / Tree-Sitter / TensorBoard

This research project compares three approaches to detecting AI-generated code: traditional machine learning classifiers, syntax-tree and graph-based representations, and deep learning embedding-based classification.

The project also forms the basis of the paper From Syntax Trees to Embeddings: A Comparative Study of AI-Generated Code Detection. A separate publication record remains deferred until its publication metadata is confirmed.

The catalogued project record lists Python, PyTorch, PyTorch-Geometric, Tree-Sitter, and TensorBoard as its core stack.

Draft status

Full write-up in progress. This page currently summarizes the catalogued project record.