Learning Moderately Input-Sensitive Functions: A Case Study in QR Code Decoding

TL;DR

This paper explores learning-based QR code decoding using Transformers, demonstrating their ability to decode beyond traditional limits and generalize across languages, revealing a novel decoding mechanism focused on data bits.

Contribution

It presents the first learning-based QR code decoder using Transformers, analyzing its ability to handle medium input sensitivity and uncovering its unique decoding strategy.

Findings

Transformers successfully decode QR codes beyond error-correction limits.

The model generalizes from English to other languages and random strings.

The decoder focuses on data bits, ignoring error-correction bits.

Abstract

The hardness of learning a function that attains a target task relates to its input-sensitivity. For example, image classification tasks are input-insensitive as minor corruptions should not affect the classification results, whereas arithmetic and symbolic computation, which have been recently attracting interest, are highly input-sensitive as each input variable connects to the computation results. This study presents the first learning-based Quick Response (QR) code decoding and investigates learning functions of medium sensitivity. Our experiments reveal that Transformers can successfully decode QR codes, even beyond the theoretical error-correction limit, by learning the structure of embedded texts. They generalize from English-rich training data to other languages and even random strings. Moreover, we observe that the Transformer-based QR decoder focuses on data bits while ignoring error-correction bits, suggesting a decoding mechanism distinct from standard QR code readers.