• Journal of the Korean Institute of Landscape Architecture
• /
• v.41 no.6
• /
• pp.1-19
• /
• 2013

This study tries to review 'Sakuteiki(作庭記)', the Book of Garden Making, compiled at the end of the 11th Century during the Heian Period of Japan, from the East-Asian perspective. 'Sakuteiki' is a Garden Theory Book, the oldest in the world as well as in Asia, and it contains the traditional knowledge of Japanese ancient garden culture, which originated from the continent(Korea and China). Traditional knowledge related to East-Asian garden culture reviewed in this paper is "Fungsu Theory"(風水, Asian traditional ecology: Fengshui in Chinese; Fusui in Japanese), stemmed from the culture to seek sound and blessed places to live in. Viewed from modern landscape architecture, the Fungsu Theory corresponds to ecology(science). The Fungsu Theory was established around the Han Dynasty of China together with the Yinyangwuxing(陰陽五行) Theory and widely used for making human residences including gardens. It was transmitted to Japan via Korea as well as through direct transaction between Japan and China. This study reinterprets garden design principles represented in Sakuteiki, which were selected in 5 key words according to the Fungsu Theory. The 5 key words for the Fungsu Theory are "the place in harmony of four guardian gods(四神相應地)", "planting trees in the four cardinal directions", "flow of Chi(氣)", "curved line and asymmetry", and "mountain is the king, water is the people". Garden design principles of "the place in harmony of four guardian gods(四神相應地)" and "planting trees in the four cardinal directions" are corresponding to "Myeongdang-ron(明堂論, Theory of propitious site)". The place in harmony of four guardian gods mentioned in Sakuteiki is a landform surrounded by the flow of water to the east, the great path to the west, the pond to the south, and the hill to the north. And the Theory originated from Zhaijing(宅經, Classic of dwelling Sites) of China. According to this principle, the city was planned and as a miniature model, the residence of the aristocrat during the Heian period was made. At the residence the location of the garden surrounded by the four gods(the flow of water, the great path, the pond, and the hill) is the Myeongdang(明堂, the propitious site: Mingtang in Chinese; Meido in Japanese). Sakuteiki explains how to substitute for the four gods by planting trees in the four cardinal directions when they were not given by nature. This way of planting originated from Zhaijing(宅經) and also goes back to Qiminyaoshu (齊民要術), compiled in the 6th Century of China. In this way of planting, the number of trees suggested in Sakuteiki is related to Hetu(河圖) and Luoshu(洛書), which are iconography of Yi(易), the philosophy of change, in ancient China. Such way of planting corresponds to that of Yongdoseo(龍圖墅, the villa based on the principle of Hetu) presented in Sanrimgyeongje (山林經濟), an encyclopedia on agriculture and living in the 17th Century of Korea. And garden design principles of "the flow of Chi(氣)", "curved line and asymmetry" is connected to "Saenggi Theory(生氣論, Theory of vitality)". Sakuteiki explains the right flow of Chi(氣) through the proper flow and the reverse flow of the garden stream and also suggests the curved line of the garden stream, asymmetric arrangement of bridges and stones in the garden, and indented shape of pond edges, which are ways of accumulating Chi(氣) and therefore lead to "Saenggi Theory" of the Fungsu Theory. The last design principle, "mountain is the king, water is the people", is related to "Hyeongguk Theory(形局論, Theory of form)" of the Fungsu Theory. Sakuteiki explains the meaning of garden through a metaphor, which views mountain as king, water as the people, and stones as king's retainers. It compares the situation in which the king governs the people with the help of his retainers to the ecological phenomena in which mountain(earth) controls water with the help of stones. This principle befits "Hyeongguk Theory(形局論, Theory of form)" of the Fungsu Theory which explains landform on the analogy of social systems, people, animals and things. As above, major garden design principles represented in Sakuteiki can be interpreted in the context of the Fungsu Theory, the traditional knowledge system in East Asia. Therefore, we can find the significance of Sakuteiki in that the wisdom of ancient garden culture in East-Asia was integrated in it, although it described the knowhow of a specific garden style in a specific period of Japan.

• Journal of Intelligence and Information Systems
• /
• v.24 no.1
• /
• pp.205-225
• /
• 2018

Convolutional Neural Network (ConvNet) is one class of the powerful Deep Neural Network that can analyze and learn hierarchies of visual features. Originally, first neural network (Neocognitron) was introduced in the 80s. At that time, the neural network was not broadly used in both industry and academic field by cause of large-scale dataset shortage and low computational power. However, after a few decades later in 2012, Krizhevsky made a breakthrough on ILSVRC-12 visual recognition competition using Convolutional Neural Network. That breakthrough revived people interest in the neural network. The success of Convolutional Neural Network is achieved with two main factors. First of them is the emergence of advanced hardware (GPUs) for sufficient parallel computation. Second is the availability of large-scale datasets such as ImageNet (ILSVRC) dataset for training. Unfortunately, many new domains are bottlenecked by these factors. For most domains, it is difficult and requires lots of effort to gather large-scale dataset to train a ConvNet. Moreover, even if we have a large-scale dataset, training ConvNet from scratch is required expensive resource and time-consuming. These two obstacles can be solved by using transfer learning. Transfer learning is a method for transferring the knowledge from a source domain to new domain. There are two major Transfer learning cases. First one is ConvNet as fixed feature extractor, and the second one is Fine-tune the ConvNet on a new dataset. In the first case, using pre-trained ConvNet (such as on ImageNet) to compute feed-forward activations of the image into the ConvNet and extract activation features from specific layers. In the second case, replacing and retraining the ConvNet classifier on the new dataset, then fine-tune the weights of the pre-trained network with the backpropagation. In this paper, we focus on using multiple ConvNet layers as a fixed feature extractor only. However, applying features with high dimensional complexity that is directly extracted from multiple ConvNet layers is still a challenging problem. We observe that features extracted from multiple ConvNet layers address the different characteristics of the image which means better representation could be obtained by finding the optimal combination of multiple ConvNet layers. Based on that observation, we propose to employ multiple ConvNet layer representations for transfer learning instead of a single ConvNet layer representation. Overall, our primary pipeline has three steps. Firstly, images from target task are given as input to ConvNet, then that image will be feed-forwarded into pre-trained AlexNet, and the activation features from three fully connected convolutional layers are extracted. Secondly, activation features of three ConvNet layers are concatenated to obtain multiple ConvNet layers representation because it will gain more information about an image. When three fully connected layer features concatenated, the occurring image representation would have 9192 (4096+4096+1000) dimension features. However, features extracted from multiple ConvNet layers are redundant and noisy since they are extracted from the same ConvNet. Thus, a third step, we will use Principal Component Analysis (PCA) to select salient features before the training phase. When salient features are obtained, the classifier can classify image more accurately, and the performance of transfer learning can be improved. To evaluate proposed method, experiments are conducted in three standard datasets (Caltech-256, VOC07, and SUN397) to compare multiple ConvNet layer representations against single ConvNet layer representation by using PCA for feature selection and dimension reduction. Our experiments demonstrated the importance of feature selection for multiple ConvNet layer representation. Moreover, our proposed approach achieved 75.6% accuracy compared to 73.9% accuracy achieved by FC7 layer on the Caltech-256 dataset, 73.1% accuracy compared to 69.2% accuracy achieved by FC8 layer on the VOC07 dataset, 52.2% accuracy compared to 48.7% accuracy achieved by FC7 layer on the SUN397 dataset. We also showed that our proposed approach achieved superior performance, 2.8%, 2.1% and 3.1% accuracy improvement on Caltech-256, VOC07, and SUN397 dataset respectively compare to existing work.