Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Bilevel Hypergraph Networks for Multi-Modal Alzheimer's Diagnosis

Angelica I. Aviles-Rivero, Chun-Wun Cheng, Zhongying Deng, Zoe Kourtzi, Carola-Bibiane Schönlieb

TL;DR

A new hypergraph framework is introduced that enables higher-order relations between multi-modal data, while utilising minimal labels, and introduces a novel strategy for generating pseudo-labels more effectively via a gradient-driven flow.

Abstract

Early detection of Alzheimer's disease's precursor stages is imperative for significantly enhancing patient outcomes and quality of life. This challenge is tackled through a semi-supervised multi-modal diagnosis framework. In particular, we introduce a new hypergraph framework that enables higher-order relations between multi-modal data, while utilising minimal labels. We first introduce a bilevel hypergraph optimisation framework that jointly learns a graph augmentation policy and a semi-supervised classifier. This dual learning strategy is hypothesised to enhance the robustness and generalisation capabilities of the model by fostering new pathways for information propagation. Secondly, we introduce a novel strategy for generating pseudo-labels more effectively via a gradient-driven flow. Our experimental results demonstrate the superior performance of our framework over current techniques in diagnosing Alzheimer's disease.

Bilevel Hypergraph Networks for Multi-Modal Alzheimer's Diagnosis

TL;DR

A new hypergraph framework is introduced that enables higher-order relations between multi-modal data, while utilising minimal labels, and introduces a novel strategy for generating pseudo-labels more effectively via a gradient-driven flow.

Abstract

Early detection of Alzheimer's disease's precursor stages is imperative for significantly enhancing patient outcomes and quality of life. This challenge is tackled through a semi-supervised multi-modal diagnosis framework. In particular, we introduce a new hypergraph framework that enables higher-order relations between multi-modal data, while utilising minimal labels. We first introduce a bilevel hypergraph optimisation framework that jointly learns a graph augmentation policy and a semi-supervised classifier. This dual learning strategy is hypothesised to enhance the robustness and generalisation capabilities of the model by fostering new pathways for information propagation. Secondly, we introduce a novel strategy for generating pseudo-labels more effectively via a gradient-driven flow. Our experimental results demonstrate the superior performance of our framework over current techniques in diagnosing Alzheimer's disease.
Paper Structure (6 sections, 3 equations, 3 figures, 2 tables)

This paper contains 6 sections, 3 equations, 3 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Methodology
  3. Multi-Modal Hypergraph Construction & Actions
  4. Bilevel Hypergraph Optimisation
  5. Experimental Results
  6. Conclusion

Figures (3)

  • Figure 1: Our Bilevel Hypergraph Learning Framework for Early Alzheimer's Disease Diagnosis. It integrates multi-modal data into a hypergraph, employing a bilevel optimisation strategy to co-learn a graph augmentation policy and a semi-supervised classifier. It introduces an innovative pseudo-label update mechanism via gradient-driven flow, aiming for enhanced learning with minimal labels.
  • Figure 2: Ablation Studies. (a) Demonstrates the impact of learned augmentations within our framework. (b) Highlights the progression of pseudo-label certainty from Epoch 1 to Epoch 150, (c)Error rates obtained using our pseudo-labels versus those generated directly from a deep network.
  • Figure 3: (a) Comparison of our technique with Dual HG across label rates, showing error rate differences. (b) Error rate comparison for Learned Policy, Heuristic, and D_Aug strategies.