SEMI-CAVA: A Causal Variational Approach to Semi-Supervised Learning.

Area of Science:

• Artificial Intelligence
• Machine Learning
• Causal Inference

Background:

• Deep learning requires large labeled datasets, which are scarce and costly in domains like medicine.
• Existing semi-supervised learning (SSL) methods have limitations in modeling complex causal relationships.
• Current causality-based SSL approaches often handle only low-dimensional data or class imbalance.

Purpose of the Study:

• To develop a causal generative model for semi-supervised learning (SSL).
• To leverage principles from causality and variational inference to improve SSL.
• To address the limitations of existing methods in modeling causal factors for SSL.

Main Methods:

• Combines principles from causality and variational inference.
• Interprets the Mixup strategy as a stochastic intervention.
• Introduces a consistency loss for coherent latent representations.
• Provides theoretical guarantees for learned representations aligning with causal factors.

Main Results:

• Achieves state-of-the-art performance on diverse medical datasets.
• Demonstrates competitive performance on standard benchmarks (CIFAR10, CIFAR100, SVHN).
• Learned latent representations align with true causal factors under stated assumptions.

Conclusions:

• The proposed causal generative model effectively enhances semi-supervised learning.
• The method reduces reliance on labeled data, particularly beneficial for medical applications.
• This work advances the integration of causality into deep learning for improved data efficiency and representation learning.