arXiv subjects

Joon Kim

Publications and source records attributed to Joon Kim.

3 recordsLinked to original sources

arXiv2026

RealDocBench: A Benchmark for Field-Level QA and Layout Understanding on Real-World Regulated Documents

Document parsing systems are increasingly deployed in high-stakes, regulated workflows such as mortgage underwriting, financial reporting, supply-chain logistics, and clinical records. Yet most public benchmarks evaluate parsers on clean academic layouts or synthetic prose, and report a single OCR or markdown-level similarity score. Such documents and metrics correlate poorly with what downstream agents actually need: the correct value for a specific field on a messy real-world page. We introduce RealDocBench, a two-track benchmark built from real regulated documents. The QA track contains 1,356 field-level questions over 581 documents spanning four domains, where each question is paired with a typed gold_dict of key-to-value answers and parsers are scored on both per-field and strict per-question accuracy. The layout track contains 1,500 human-verified page images annotated with COCO-style bounding boxes under a nine-class public taxonomy, scored with a Hungarian matcher that includes adjacency-aware split/merge recovery. We evaluate eighteen systems, spanning commercial parsing APIs, general-purpose VLMs, and open-source OCR models, under a uniform extraction-and-scoring protocol, and report accuracy alongside per-page cost and cache-busted latency. RealDocBench exposes a wide performance spread that single-number benchmarks hide, a persistently hard medical sub-domain, and sharp cost/latency trade-offs across operating points. We release the datasets, parser adapters, and evaluation harness to support reproducible, field-level comparison of document parsing systems.

cs.CV↗

arXiv2025

Bit-Flipping Attack Exploration and Countermeasure in 5G Network

5G communication technology has become a vital component in a wide range of applications due to its unique advantages such as high data rate and low latency. While much of the existing research has focused on optimizing its efficiency and performance, security considerations have not received comparable attention, potentially leaving critical vulnerabilities unexplored. In this work, we investigate the vulnerability of 5G systems to bit-flipping attacks, which is an integrity attack where an adversary intercepts 5G network traffic and modifies specific fields of an encrypted message without decryption, thus mutating the message while remaining valid to the receiver. Notably, these attacks do not require the attacker to know the plaintext, and only the semantic meaning or position of certain fields would be enough to effect targeted modifications. We conduct our analysis on OpenAirInterface (OAI), an open-source 5G platform that follows the 3GPP Technical Specifications, to rigorously test the real-world feasibility and impact of bit-flipping attacks under current 5G encryption mechanisms. Finally, we propose a keystream-based shuffling defense mechanism to mitigate the effect of such attacks by raising the difficulty of manipulating specific encrypted fields, while introducing no additional communication overhead compared to the NAS Integrity Algorithm (NIA) in 5G. Our findings reveal that enhancements to 5G security are needed to better protect against attacks that alter data during transmission at the network level.

cs.CR↗

arXiv2024

Random Gradient Masking as a Defensive Measure to Deep Leakage in Federated Learning

Federated Learning(FL), in theory, preserves privacy of individual clients' data while producing quality machine learning models. However, attacks such as Deep Leakage from Gradients(DLG) severely question the practicality of FL. In this paper, we empirically evaluate the efficacy of four defensive methods against DLG: Masking, Clipping, Pruning, and Noising. Masking, while only previously studied as a way to compress information during parameter transfer, shows surprisingly robust defensive utility when compared to the other three established methods. Our experimentation is two-fold. We first evaluate the minimum hyperparameter threshold for each method across MNIST, CIFAR-10, and lfw datasets. Then, we train FL clients with each method and their minimum threshold values to investigate the trade-off between DLG defense and training performance. Results reveal that Masking and Clipping show near to none degradation in performance while obfuscating enough information to effectively defend against DLG.

cs.LG↗