Jingwen Huang | Searcharxiv

arXiv2025

A Kinematic and Kinetic Dataset of Lower Limb Joints During Obstacle Crossing in Healthy Young Adults

Obstacle crossing is an essential component of human locomotion, particularly for individuals with lower limb amputations who face elevated risks of imbalance and falls. While prior studies have explored this task, they often lack a comprehensive examination of kinematic and kinetic changes throughout the entire gait cycle across varying obstacle heights. This study creates a novel dataset collected from ten healthy adults performing obstacle crossing at four different heights (7.5 cm, 15 cm, 22.5 cm, and 30 cm). Kinematic and kinetic data (angles and torques of hip, knee, and ankle) were recorded and analyzed. Results indicate that increased obstacle height leads to a longer swing phase and significant increases in both hip and knee joint angles (1.5* and 1.0*, respectively) and torques. In contrast, ankle joint angles and moments exhibited minimal variation across obstacle heights, indicating a relatively consistent movement strategy at the ankle. Furthermore, significant asymmetries were observed between the dominant and non-dominant foot: the dominant foot demonstrated larger hip and knee joint angles and more consistent ankle behavior, reflecting greater coordination. These findings offer valuable biomechanical insights for improving fall prevention strategies and informing the design of assistive devices such as prostheses and exoskeletons.

physics.med-ph↗

arXiv2025

Agentic Privacy-Preserving Machine Learning

Privacy-preserving machine learning (PPML) is critical to ensure data privacy in AI. Over the past few years, the community has proposed a wide range of provably secure PPML schemes that rely on various cryptography primitives. However, when it comes to large language models (LLMs) with billions of parameters, the efficiency of PPML is everything but acceptable. For instance, the state-of-the-art solution for confidential LLM inference represents at least 10,000-fold slower performance compared to plaintext inference. The performance gap is even larger when the context length increases. In this position paper, we propose a novel framework named Agentic-PPML to make PPML in LLMs practical. Our key insight is to employ a general-purpose LLM for intent understanding and delegate cryptographically secure inference to specialized models trained on vertical domains. By modularly separating language intent parsing - which typically involves little or no sensitive information - from privacy-critical computation, Agentic-PPML completely eliminates the need for the LLMs to process the encrypted prompts, enabling practical deployment of privacy-preserving LLM-centric services.

cs.CR↗