Tang and Song Ink Painting Context to Taoist Philosophy and Mikels Emotion Classification

Abstract

Tang and Song Ink Painting Context to Taoist Philosophy and Mikels Emotion Classification QINYAN ZHANG Dept. of School of Design The Graduate School Hanyang University In the long river of art, each country's artistic development boasts unique characteristics. Ink painting, stemming from Chinese tradition, serves as a vessel for Chinese culture and is a microcosm of Chinese cultural and philosophical existence, making it a valuable research category. The evolution of ink painting as an artistic form is indelibly linked to the influence of traditional Chinese culture, with the Taoist philosophy espoused by Laozhuang (老庄) being particularly significant. As a principal school of ancient Chinese philosophy, Taoism emphasizes harmony and balance between humanity and nature, advocates the principle of effortless action, and encourages emotional introspection and expression. At the heart of Chinese philosophy lies the concept of "Tao"(道) which explores the interrelations between entities and nature and the inherent laws of things, with humanity as the intermediary. This philosophical thought is mirrored in numerous ink paintings. This paper, taking the "Tao" as its foundation, examines ancient ink paintings from the Tang and Song dynasties (618-1279), analyzing the formation of ink painting styles and the aesthetic and symbolic elements within artistic practice. It discusses the philosophical tenets manifest in ink paintings, which not only express the artist's inner world but also act as conduits for ideation. Emotion is the art's core and the primary means by which viewers gain aesthetic experience and artistic insight. An exploration into the emotive aspect of ink painting enhances public understanding and preservation of Chinese painting heritage and promotes the advancement of modern ink painting. With the proliferation of the Internet, vast amounts of data are generated daily, and deep learning aids in uncovering patterns and information within this data. Emotional computing has emerged as an important field in artificial intelligence, and the fusion of traditional art with modern technology is a burgeoning trend. Information on the Internet has evolved from text to diverse modalities, including images, audio, and video. As a quintessential aspect of Chinese culture and art, ink painting is gaining international exposure through digitalization and electronic dissemination by various platforms. Images, with their rich informational content and efficient transmission, have become vital in self-expression and understanding others in daily life. Ink painting, as a visual art form, encapsulates Taoist philosophical thought and underscores the salience of human emotions and nature. The artist often conveys feelings through techniques such as line, ink shade, and composition. For instance, a landscape painting may depict grandeur and passion with rugged mountains and turbulent rivers, while a flora and fauna painting might convey tranquility and harmony with subdued colors and gentle strokes. Image Emotion Computing quantitatively assesses human emotions and moods through computer technology, utilizing algorithms and models to extract and interpret the emotional content of images, thus enabling precise recognition and comprehension of image emotions. This paper emulates the human brain's emotional processing during image perception by designing a network that teaches machines to 'feel' in a human-like manner. By adopting Taoist philosophical concepts as the emotional scale and integrating them with Mikels' emotion classification, This paper uses deep learning technology to quantify emotions in ink works, such as joy, sadness, anger, etc., allowing the public to deeply understand the works' connotations and emotional expressions and how artistic philosophy conveys emotions. Deep learning processes the lines, colors, and elements of ink paintings to identify the emotions and sentiments contained within, offering valuable insights for the evolution of modern Chinese ink paintings. Furthermore, the sentiment analysis of ink painting images has broad applications, including in social media sentiment analysis, intelligent advertising, user sentiment assessment, intelligent interaction and experiences, and in psychological and medical research. This paper comprises nine sections that elaborate on the topic (Figure 1). highlighting the macroscopic and microstructural connections of Chinese Taoist philosophy to ink painting. At the macro level, the connection between nature and humanity is explored, and at the micro level, ink painting is deconstructed into its basic elements: human thoughts and emotions, materials and techniques, and formal symbols. This forms a cyclical chain: nature gives rise to humanity, humanity creates ink painting, and ink painting, through materials and techniques, expresses itself as formal symbols that reflect both humanity and nature.Delves into the stylistic expressions of Tang and Song ink paintings across different genres- landscape, humanistic, and flora and fauna- and examines the works of notable painters in each category，extracts the characteristics of vitality, thoughtfulness, personal expression, and line variety from classical paintings. Explain the position and role of Taoist philosophical thinking in Chinese culture from historical allusions. Summarize the basic connotation of Chinese Taoist philosophy and how "Tao" came into being, and explain the influence of Taoist philosophy in art. Taoist philosophy is expressed in ink painting in four ways, which can be summarized as geometric philosophy, color philosophy, white space philosophy and natural philosophy. Analyzing specific paintings of the Tang and Song dynasties from these four aspects, it distills how ink painting embodies philosophy. In the second half of the paper explains the fundamental concepts and significance of emotional image computing. It explores the definition and categorization of emotions and justifies the use of Mikels' 8-category emotion model in this thesis. Summarizes the behaviorist and cognitivist views in psychology to comprehend emotion genesis and presents the integration of deep learning network technology for art-based emotional information analysis and understanding. Human emotion generation mainly consists of three steps: Stimuli, Cognition and Feedback. Human emotions are generated from specific emotional stimuli rather than the whole image, which makes stimulus selection crucial for image emotion analysis. Therefore, unlike directly extracting features from the whole ink painting, this paper will select emotional stimuli from ink paintings, and then further extract emotional features in these stimuli. Before starting the emotion modeling task, it is important to clarify the importance of dataset specification first. In this thesis, the data of Tang and Song ink paintings are collected and processed, which are divided into training set and test set, and the data ratio of training set and validation set is 9:1, in which the ink paintings are divided into two categories of Positive and Negative and the distribution ratio of the two categories is 4:6, which ensures that the training set contains a diversity of sentiment samples. Finally, the image data are preprocessed and labeled for subsequent analysis of sentiment image computation. Present the results of the analysis in a visual form such as a chart. For the construction of the model architecture, the author chose the most suitable network architecture for sentiment analysis of ink drawings, MobileNet, which is a network model optimized specifically for mobile devices and embedded platforms. Its lightweight structure and low parameter count make it possible to realize real-time sentiment analysis with limited computational resources. Considering that ink painting sentiment analysis may need to be performed in real-time on mobile devices, the use of MobileNet can help the ink painting sentiment classification model to run the task efficiently in such an environment. This paper also details the structures used in the MobileNet network architecture in terms of the underlying logic, convolutional layers, depth-separable convolution, Bottleneck block, Squeeze-and-Excitation Module, Average Pooling, hard-sigmoid function and hard- switch function . In terms of model training and application, emphasizing the importance of training in deep learning. It involves using labeled data to fine-tune a neural network model to better align with the task and minimize the loss function value. The deployed program can output sentiment analysis data after inputting a picture. Through the training of the model, it is verified that the ink drawing sentiment program analysis accuracy reaches 0.85, which achieves good results. Outputs are categorized as hard or soft; hard outputs provide standard sentiment analysis and determine sentiment polarity, leading to the generation of evaluative text based on the polarity, whereas soft outputs present raw sentiment analysis data. By default, hard outputs are used, and a radar chart is generated from the sentiment data. In order to reduce the burden on the user's device and to facilitate the user's ability to access the ink drawing sentiment classification model from anywhere via the Internet, the author deployed the application as a web page. The IP address is: http://124.222.246.30/.(Server duration is 2023.10.31-2024.10.25). In the final part of this thesis concludes with insights into how the emotions evoked by artistic elements can be applied to contemporary ink painting creation. The relationship between image emotion analysis and art creation offers constructive insights for ink painting's developmental trajectory, contributing intrinsic meaning and value to the innovation and progression of modern ink painting. Figure 1. Thesis Structure Cha