• 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.

• Membrane Journal
• /
• v.26 no.4
• /
• pp.239-252
• /
• 2016

Two-dimensional materials offer unique characteristics for membrane applications to water technology. With its atomic thickness, availability and stackability, graphene in particular is attracting attention in the research and industrial communities. Here, we present a brief overview of the recent research activities in this rising topic with bringing two membrane architecture into focus. Pristine graphene in single- and polycrystallinity poses a unique diffusion barrier property for most of chemical species at broad ambient conditions. If well designed and controlled, physical and chemical perforation can turn this barrier layer to a thinnest feasible membrane that permits ultimate permeation at given pore sizes. For subcontinuum pores, both molecular dynamics simulations and experiments predict potential salt rejection to envisage a seawater desalination application. Another novel membrane architecture is a stack of individual layers of 2D materials. When graphene-based platelets are chemically modified and stacked, the interplanar spacing forms a narrow transport pathway capable of separation of solvated ions from pure water. Bearing unbeknownst permeance and selectivity, both membrane architecture - ultrathin porous graphene and stacked platelets - offer a promising prospect for new extraordinary membranes for water technology applications.

• The Research Journal of the Costume Culture
• /
• v.18 no.2
• /
• pp.261-276
• /
• 2010

The analysis of an art trend in the principle dimension starts by observing the object of work in the perspective of formative composition and recognizing it as a universal system. It can be said that it is consistent with an interpretation method for a form theory of formal history by Heinrich W$\ddot{o}$lfflin, a leading form critic in art criticism. Hence, the purpose of this study was to find out what are the formative principles in Renaissance Classicism as a design principle to be applicable to modern fashion by reviewing the formative characteristics of Renaissance Classicism Architecture with which W$\ddot{o}$lfflin directly dealt. As for the theoretical literature review, I used W$\ddot{o}$lfflin's theoretical framework and looked at the Renaissance Classicism Architecture that he studied and examined the possibility of utilizing his theory as a layout principle and the characteristics. As for analysis of design cases, I applied the aforementioned architecture layout principle to modern fashion and conducted case study analysis to delve into distinctive layout principles found in fashion. The study showed that the Renaissance Classicism Architectural Style is marked by linearity, planarity, closing and multiple unity: linearity was expressed in the observation form in fixed frontal view and an emphasis on a tangible silhouette homeogenous and definite line structures; planarity was achieved in the form of paralleled layers of frontal view element, planarity style, and identical and proportional repetition of various sizes.; closing signified the pursuit of complete and clear regularity, and architecture developed in a constructive phase through organizational inevitability and absolute invariability.; multiple unity was expressed in self-completedness and independent parallel of discrete forms and harmony of emphasized individual elements in a totality. Applying these layout characteristics of the Renaissance Classicism Architectural style and to see their individual expressive features, I found out that in adopting layout principles of the Renaissance Classicism Architecture to modern fashion, it turned out to be an emphasis of individual silhouettes, a flattened space, completed objects, organic harmony among independent parts: the emphasis of individual silhouettes was expressed in individual definitiveness of formative lines of clothes in accordance with body joints and an emphasis on formative lines of clothes; the flattened space was marked by single layer structure, planarity of elements of clothes, and listing arrangement by appropriate proportion.; the completedness of the objects was expressed by the stationary state where overall image is fixed, the construction of homogeneous and complete space, and absolute inevitability of internal layout in proportion; lastly, organic harmony of independent parts was stressed in independent completedness of each detail, and organic harmony of the whole. The expressive features would lead to a unique expression style of linear emphasis, proportion, constructive forms, and two-dimensional arrangement. The meaning of this study is follows: The characteristics of art school of thought are given shape by appling & analysing the architectural layout principles of historical art school of thought to modern fashion in the view point of formal construction dimension. The applied possibility of historical art school of thought as the source of inspiration about the fashion design is extended.