Ensemble Learning with Sparse Hypercolumns

2026-03-06

AI Summary

This work addresses the high computational cost and susceptibility to overfitting of dense hypercolumn representations in image segmentation, particularly under limited training data. To mitigate these issues, the authors propose a sparse hypercolumn representation method that reduces dimensionality through hierarchical subsampling and integrates logistic regression with stacked/voting ensemble strategies for pixel-wise classification. As the first systematic exploration of combining sparse hypercolumns with ensemble learning, the approach achieves a mean Dice coefficient of 0.66 on brain tumor segmentation using only 20 training samples and a 10% subsampling rate—outperforming a U-Net baseline by 24.53% (p = 3.07 × 10⁻¹¹). This demonstrates significant improvements in both performance and efficiency for few-shot segmentation tasks.

Technology Category

• Computer Vision: Segmentation
• Machine Learning: Ensemble Methods
• Cognitive Modeling & Cognitive Systems: Neural Spike Coding

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets

Abstract

Directly inspired by findings in biological vision, high-dimensional hypercolumns are feature vectors built by concatenating multi-scale activations of convolutional neural networks for a single image pixel location. Together with powerful classifiers, they can be used for image segmentation i.e. pixel classification. However, in practice, there are only very few works dedicated to the use of hypercolumns. One reason is the computational complexity of processing concatenated dense hypercolumns that grows linearly with the size $N$ of the training set. In this work, we address this challenge by applying stratified subsampling to the VGG16 based hypercolumns. Furthermore, we investigate the performance of ensemble learning on sparse hypercolumns. Our experiments on a brain tumor dataset show that stacking and voting ensembles deliver competitive performance, but in the extreme low-shot case of $N \leq 20$, a simple Logistic Regression classifier is the most effective method. For 10% stratified subsampling rate, our best average Dice score is 0.66 for $N=20$. This is a statistically significant improvement of 24.53% over the standard multi-scale UNet baseline ($p$-value = $[3.07e-11]$, Wilcoxon signed-rank test), which is less effective due to overfitting.

Problem

• hypercolumns
• computational complexity
• image segmentation
• ensemble learning
• sparse representation

Innovation

• Sparse Hypercolumns
• Ensemble Learning
• Stratified Subsampling
• Low-shot Segmentation
• VGG16