Effectiveness of Deep Image Embedding Clustering Methods on Tabular Data
This study highlights a data-centric gap for researchers and practitioners in machine learning, showing that incremental customization is needed for deep methods to compete with conventional approaches on tabular data.
The paper tackled the problem of applying deep image embedding clustering methods to tabular data, finding that a traditional clustering method outperformed most deep learning baselines, ranking second out of eight methods tested on seven tabular datasets.
Deep learning methods in the literature are commonly benchmarked on image data sets, which may not be suitable or effective baselines for non-image tabular data. In this paper, we take a data-centric view to perform one of the first studies on deep embedding clustering of tabular data. Eight clustering and state-of-the-art embedding clustering methods proposed for image data sets are tested on seven tabular data sets. Our results reveal that a traditional clustering method ranks second out of eight methods and is superior to most deep embedding clustering baselines. Our observation aligns with the literature that conventional machine learning of tabular data is still a robust approach against deep learning. Therefore, state-of-the-art embedding clustering methods should consider data-centric customization of learning architectures to become competitive baselines for tabular data.