arXiv subjects

Cynthia Huang

Publications and source records attributed to Cynthia Huang.

2 recordsLinked to original sources

arXiv2026

Digital State Capacity

Digital State Capacity is the ability of governments to deploy ICT infrastructure and information systems to implement policy. This paper introduces a new measure of government ICT capacity based on an observable stock of deployable public-sector network infrastructure: public IPv4 address space held by government organisations. These address holdings are key inputs into digital administration because they support internet-facing systems, networked information exchange, and coordination across agencies and functions. The core panel covers approximately 150,000 country-entity records classified as government across more than 150 countries from 2019 to 2024 and can be disaggregated by administrative level and government function. In the 2019 to 2024 Admin-1 panel, government IP holdings are observed in 1,681 subnational regions across all years. We validate the measure at the crosscountry and subnational levels and apply it to government tasks related to corruption control and vaccination rollout. In illustrative country-year analysis, higher Digital State Capacity is associated with higher-quality governance and publicservice outcomes in the expected directions, including lower measured corruption and higher vaccination coverage. These associations are descriptive; they demonstrate the empirical relevance of the measure and are not causal estimates.

econ.GN↗

arXiv2023

Approximating Human-Like Few-shot Learning with GPT-based Compression

In this work, we conceptualize the learning process as information compression. We seek to equip generative pre-trained models with human-like learning capabilities that enable data compression during inference. We present a novel approach that utilizes the Generative Pre-trained Transformer (GPT) to approximate Kolmogorov complexity, with the aim of estimating the optimal Information Distance for few-shot learning. We first propose using GPT as a prior for lossless text compression, achieving a noteworthy compression ratio. Experiment with LLAMA2-7B backbone achieves a compression ratio of 15.5 on enwik9. We justify the pre-training objective of GPT models by demonstrating its equivalence to the compression length, and, consequently, its ability to approximate the information distance for texts. Leveraging the approximated information distance, our method allows the direct application of GPT models in quantitative text similarity measurements. Experiment results show that our method overall achieves superior performance compared to embedding and prompt baselines on challenging NLP tasks, including semantic similarity, zero and one-shot text classification, and zero-shot text ranking.

cs.AI↗