• 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.

• Korean Journal of Heritage: History & Science
• /
• v.37
• /
• pp.285-307
• /
• 2004

Synthesize and examine petrological characteristic and geochemical characteristic by weathering formation of rock and progress of weathering laying stress on stone cultural properties of Unjusa temple of Chonnam Hwasun county site in this research. Examine closely weathering element that influence mechanical, chemical, mineralogical and physical weathering of rocks that accomplish stone cultural properties and these do quantification, wish to utilize by a basic knowledge for conservation scientific research of stone cultural properties by these result. Enforced component analysis of rock and mineralogical survey about 18 samples (pyroclastic tuff; 7, ash tuff; 4, granite ; 4, granitic gneiss; 3) all to search petrological characteristic and geochemical characteristic by weathering of Unjusa temple precinct stone cultural properties and recorded deterioration degree about each stone cultural properties observing naked eye. Major rock that constitution Unjusa temple one great geological features has strike of N30-40W and dip of 10-20NE being pyroclastic tuff. This pyroclastic tuff is ranging very extensively laying center on Unjusa temple and stone cultural properties of precinct is modeled by this pyroclastic tuff. Stone cultural propertieses of present Unjusa temple precinct are accomplishing structural imbalance with serious crack, and because weathering of rock with serious biological pollution is gone fairly, rubble break away and weathering and deterioration phenomenon such as fall off of a particle of mineral are appearing extremely. Also, a piece of iron and cement mortar of stone cultural properties everywhere are forming precipitate of reddish brown and light gray being oxidized. About these stone cultural properties, most stone cultural propertieses show SD(severe damage) to MD(moderate damage) as result that record Deterioration degree. X-ray diffraction analysis result samples of each rock are consisted of mineral of quartz, orthoclase,plagioclase, calcite, magnetite etc. Quartz and feldspar alterated extremely in a microscopic analysis, and biotite that show crystalline form of anhedral shows state that become chloritization that is secondary weathering mineral being weathered. Also, see that show iron precipitate of reddish brown to crack zone of tuff everywhere preview rock that weathering is gone deep. Tuffs that accomplish stone cultural properties of study area is illustrated to field of Subalkaline and Peraluminous, $SiO_2$(wt.%) extent of samples pyroclastic tuff 70.08-73.69, ash tuff extent of 70.26-78.42 show. In calculate Chemical Index of Alteration(CIA) and Weathering Potential Index(WPI) about major elements extent of CIA pyroclastic tuff 55.05-60.75, ash tuff 52.10-58.70, granite 49.49-51.06 granitic gneiss shows value of 53.25-67.14 and these have high value gneiss and tuffs. WPI previews that is see as thing which is illustrated being approximated in 0 lines and 0 lines low samples of tuffs and gneiss is receiving esaily weathering process as appear in CIA. As clay mineral of smectite, zeolite that is secondary weathering produce of rock as result that pick powdering of rock and clothing material of stone cultural properties observed by scanning electron micrographs (SEM). And roots of lichen and spore of hyphae that is weathering element are observed together. This rock deep organism being coating to add mechanical weathering process of stone cultural properties do, and is assumed that change the clay mineral is gone fairly in stone cultural properties with these. As the weathering of rocks is under a serious condition, the damage by the natural environment such as rain, wind, trees and the ground is accelerated. As a counter-measure, the first necessary thing is to build the ground environment about protecting water invasion by making the drainage and checking the surrounding environment. The second thing are building hardening and extirpation process that strengthens the rock, dealing biologically by reducing lichens, and sticking crevice part restoration using synthetic resin. Moreover, it is assumed to be desirable to build the protection facility that can block wind, sunlight, and rain which are the cause of the weathering, and that goes well with the surrounding environment.