A novel network for classification of cuneiform tablet metadata

TL;DR

This paper introduces a new convolution-inspired neural network architecture for classifying cuneiform tablet metadata, effectively handling limited data and high-resolution point clouds, outperforming transformer-based models.

Contribution

The paper proposes a novel network design tailored for cuneiform tablet classification, combining local and global information processing, and demonstrating superior performance over existing transformer-based methods.

Findings

The proposed network outperforms Point-BERT in classification accuracy.

The architecture effectively integrates local neighbor and global features.

Source code and datasets will be publicly released.

Abstract

In this paper, we present a network structure for classifying metadata of cuneiform tablets. The problem is of practical importance, as the size of the existing corpus far exceeds the number of experts available to analyze it. But the task is made difficult by the combination of limited annotated datasets and the high-resolution point-cloud representation of each tablet. To address this, we develop a convolution-inspired architecture that gradually down-scales the point cloud while integrating local neighbor information. The final down-scaled point cloud is then processed by computing neighbors in the feature space to include global information. Our method is compared with the state-of-the-art transformer-based network Point-BERT, and consistently obtains the best performance. Source code and datasets will be released at publication.