Skip to content

Architecture Modification / Research · Python

soft-tree-gnn-for-elliptic--

What it is

This repository implements a memory-efficient GAT variant (HeadChunkedGATConv) and combines it with a differentiable soft decision tree in an end-to-end trainable model for AML detection on the Elliptic dataset. It includes training loops, ablation studies, continual learning post-drift, and explainability features.

Scored at5a64b0c19 Nov 2026Changes since

1 newer commit is not scored yet. They count once the owner rescans. How merit is scored

Distribution based on AI work

1 / 2
Machine Learning: 56.3%MLOps: 25.0%Data Science: 18.7%
  • Machine Learning56%
  • MLOps25%
  • Data Science19%

Each field's share of the project's AI work, as scoring counts it. It ranks in the field it is most about.

What the judge found

Read from the project's code at its scored commit.

Strengths4

  • Novel memory-efficient GATConv implementation (HeadChunkedGATConv) that reduces VRAM usage.
  • End-to-end trainable GNN+soft decision tree architecture providing interpretability via decision paths.
  • Comprehensive ablation studies and continual learning experiments validating design choices.
  • Application to a real-world AML detection task with explainability features.

Limitations

The owner keeps this project's limitations private.

Around it on the Machine Learning board

Whole board
  1. 17masld-fib4ML Application · Python64.8/100AdvancedMohammad Ahmadullah Khan
  2. 18camera505ML Application · Python64.6/100Advancedvlad andrei
  3. 19soft-tree-gnn-for-elliptic--Architecture Modification / Research · Python63.6/100AdvancedMaks
  4. 20, shared rankocr-livive-netEducational Implementation · JavaScript63.4/100AdvancedJarad Peckenpaugh
  5. 20, shared rankUOMindLLMML Application · Python63.4/100AdvancedDaniel Strandt

More from Maks