Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
This paper investigates the application of an importance-scoring metric to multi-head transformer models specifically for tabular data, an area that has received limited attention compared to vision and language tasks. By analyzing 40 diverse tabular datasets, the authors demonstrate that removing heads with low importance scores leads to minimal performance degradation, while the removal of high-importance heads significantly impacts classification accuracy. The findings reveal that important attention heads are distributed across layers without consistent trends, highlighting the unique challenges of applying transformers to tabular data.
Removing the least important attention heads from transformers can preserve performance in tabular data tasks, challenging conventional wisdom about model architecture optimization.
Computationally demanding and opaque deep learning models can be better understood and optimized by analyzing how they transform data. While deep transformers have been widely studied in computer vision and natural language processing, their application in tabular data remains relatively underexplored. This paper presents one of the first applications of an importance-scoring metric to interpret multi-head transformer models in learning from tabular data. Experiments conducted on 40 diverse tabular datasets demonstrate robustness to head drops based on the proposed head importance score. In 72.5\% of experimental examples, the model remains most resilient to performance drops when heads with the lowest importance scores are gradually removed. In contrast, removing the most important attention head first results in the greatest reduction in classification performance. A closer look at individual head importance scores across six attention layers reveals that important heads are scattered across layers, with no consistent layer-specific trends. In contrast to the image and language domains, the importance of individual attention heads varies considerably across tabular datasets with different schemas and feature spaces. The proposed importance score can improve efficiency and redundancy within transformer architectures. We make the source code for measuring the importance of individual attention heads publicly available.