arXiv subjects

Ning Zou

Publications and source records attributed to Ning Zou.

3 recordsLinked to original sources

arXiv2025

Knowledge Graph for Intelligent Generation of Artistic Image Creation: Constructing a New Annotation Hierarchy

Our study aims to establish a unified, systematic, and referable knowledge framework for the annotation of art image datasets, addressing issues of ambiguous definitions and inconsistent results caused by the lack of common standards during the annotation process. To achieve this goal, a hierarchical and systematic art image knowledge graph was constructed. It was developed based on the composition principles of art images, incorporating the Structured Theory of Visual Knowledge proposed by Academician Yunhe Pan in On Visual Knowledge-which states that visual knowledge must achieve precise expression of spatial forms and dynamic relationships through "prototype-category" and "hierarchical structure". Through in-depth review of Chinese and Western art theories and pioneering integration of the Chinese cultural perspective, this graph took shape. The core visual language of art images was deconstructed by this knowledge graph. Meanwhile, the unique spatial theory and symbolic system of Chinese painting were compared with and supplemented by Western art theories. This graph converts qualitative artistic concepts into a clear structured framework. It not only conforms to the cognitive law that "visual knowledge takes precedence over verbal knowledge" in humans but also provides an interpretable and inferential visual knowledge foundation for AI art generation and cross-cultural art analysis. It ensures the high quality and consistency of annotated data, thus offering key support for art intelligence research in the AI 2.0 era.

cs.HC↗

arXiv2024

Effects of Different Prompts on the Quality of GPT-4 Responses to Dementia Care Questions

Evidence suggests that different prompts lead large language models (LLMs) to generate responses with varying quality. Yet, little is known about prompts' effects on response quality in healthcare domains. In this exploratory study, we address this gap, focusing on a specific healthcare domain: dementia caregiving. We first developed an innovative prompt template with three components: (1) system prompts (SPs) featuring 4 different roles; (2) an initialization prompt; and (3) task prompts (TPs) specifying different levels of details, totaling 12 prompt combinations. Next, we selected 3 social media posts containing complicated, real-world questions about dementia caregivers' challenges in 3 areas: memory loss and confusion, aggression, and driving. We then entered these posts into GPT-4, with our 12 prompts, to generate 12 responses per post, totaling 36 responses. We compared the word count of the 36 responses to explore potential differences in response length. Two experienced dementia care clinicians on our team assessed the response quality using a rating scale with 5 quality indicators: factual, interpretation, application, synthesis, and comprehensiveness (scoring range: 0-5; higher scores indicate higher quality).

cs.CL↗

arXiv2020

User independent Emotion Recognition with Residual Signal-Image Network

User independent emotion recognition with large scale physiological signals is a tough problem. There exist many advanced methods but they are conducted under relatively small datasets with dozens of subjects. Here, we propose Res-SIN, a novel end-to-end framework using Electrodermal Activity(EDA) signal images to classify human emotion. We first apply convex optimization-based EDA (cvxEDA) to decompose signals and mine the static and dynamic emotion changes. Then, we transform decomposed signals to images so that they can be effectively processed by CNN frameworks. The Res-SIN combines individual emotion features and external emotion benchmarks to accelerate convergence. We evaluate our approach on the PMEmo dataset, the currently largest emotional dataset containing music and EDA signals. To the best of author's knowledge, our method is the first attempt to classify large scale subject-independent emotion with 7962 pieces of EDA signals from 457 subjects. Experimental results demonstrate the reliability of our model and the binary classification accuracy of 73.65% and 73.43% on arousal and valence dimension can be used as a baseline.

cs.CV↗