• Economic and Environmental Geology
• /
• v.50 no.3
• /
• pp.215-224
• /
• 2017

Arsenic (As) is known to be the most toxic element and frequently detected in groundwater environment. Inorganic As exists as arsenite [As(III)] and arsenate [As(V)] in reduced and oxidized environments, respectively. It has been reported that the toxicity of arsenite is much higher than that of arsenate and furthermore arsenite shows relatively higher mobility in aqueous environments. For this reason, there have been numerous researches on the process for oxidation of arsenite to arsenate to reduce the toxicity of arsenic. In particular, photooxidation has been considered to be simple, economical, and efficient to attain such goal. This study was conducted to evaluate the applicability of naturally-occurring goethite as a photocatalyst to substitute for $TiO_2$ which has been mostly used in the photooxidation processes so far. In addition, the effects of several factors on the overall performance of arsenite photocatalytic oxidation process were evaluated. The results show that the efficiency of the process was affected by total concentration of dissolved cations rather than by the kind of those cations and also the relatively higher pH conditions seemed to be more favorable to the process. In the case of coexistence of arsenite and arsenate, the removal tendency by adsorption onto goethite appeared to be different between arsenite and arsenate due to their different affinities with goethite, but any effect on the photocatalytic oxidation of arsenite was not observed. In terms of effect of humic acid on the process, it is likely that the higher concentration of humic acid reduced the overall performance of the arsenite photocatalytic oxidation as a result of competing interaction of activated oxygen species, such as hydroxyl and superoxide radicals, with arsenite and humic acid. In addition, it is revealed that the injection of oxygen gas improved the process because oxygen contributes to arsenite oxidation as an electron acceptor. Based on the results of the study, consequently, the photocatalytic oxidation of aqueous arsenite using goethite seems to be greatly feasible with the optimization of process.

• Clean Technology
• /
• v.27 no.4
• /
• pp.332-340
• /
• 2021

With the recent reinforcement of emission standards, it is necessary to make efforts to reduce NOx from air pollutant-emitting workplaces. The NOx reduction method mainly used in industrial facilities is selective catalytic reduction (SCR), and the most commercial SCR catalyst is the ceramic honeycomb catalyst. This study was carried out to reduce the NOx emitted from steel plants by applying De-NOx catalyst coated on metallic monolith. The De-NOx catalyst was synthesized through the optimized coating technique, and the coated catalyst was uniformly and strongly adhered onto the surface of the metallic monolith according to the air jet erosion and bending test. Due to the good thermal conductivity of metallic monolith, the De-NOx catalyst coated on metallic monolith showed good De-NOx efficiency at low temperatures (200 ~ 250 ℃). In addition, the optimal amount of catalyst coating on the metallic monolith surface was confirmed for the design of an economical catalyst. Based on these results, the De-NOx catalyst of commercial grade size was tested in a semi-pilot De-NOx performance facility under a simulated gas similar to the exhaust gas emitted from a steel plant. Even at a low temperature (200 ℃), it showed excellent performance satisfying the emission standard (less than 60 ppm). Therefore, the De-NOx catalyst coated metallic monolith has good physical and chemical properties and showed a good De-NOx efficiency even with the minimum amount of catalyst. Additionally, it was possible to compact and downsize the SCR reactor through the application of a high-density cell. Therefore, we suggest that the proposed De-NOx catalyst coated metallic monolith may be a good alternative De-NOx catalyst for industrial uses such as steel plants, thermal power plants, incineration plants ships, and construction machinery.

• Journal of Wetlands Research
• /
• v.20 no.4
• /
• pp.424-431
• /
• 2018

Stormwater runoff containing considerable amounts of pollutants such as particulates, organics, nutrients, and heavy metals contaminate natural bodies of water. At present, best management practices (BMP) intended to reduce the volume and treat pollutants from stormwater runoff were devised to serve as cost-effective measures of stormwater management. However, improper design and lack of proper maintenance can lead to degradation of the facility, making it unable to perform its intended function. This study evaluated an infiltration trench (IT) that went through a series of maintenance operations. 41 monitored rainfall events from 2009 to 2016 were used to evaluate the pollutant removal capabilities of the IT. Assessment of the water quality and hydrological data revealed that the inflow volume was the most relative factor affecting the unit pollutant loads (UPL) entering the facility. Seasonal variations also affected the pollutant removal capabilities of the IT. During the summer season, the increased rainfall depths and runoff volumes diminished the pollutant removal efficiency (RE) of the facility due to increased volumes that washed off larger pollutant loads and caused the IT to overflow. Moreover, the system also exhibited reduced pollutant RE for the winter season due to frozen media layers and chemical-related mechanisms impacted by the low winter temperature. Maintenance operations also posed considerable effects of the performance of the IT. During the first two years of operation, the IT exhibited a decrease in pollutant RE due to aging and lack of proper maintenance. However, some events also showed reduced pollutant RE succeeding the maintenance as a result of disturbed sediments that were not removed from the geotextile. Ultimately, the presented effects of maintenance operations in relation to the pollutant RE of the system may lead to the optimization of maintenance schedules and procedures for BMP of same structure.

• Journal of Intelligence and Information Systems
• /
• v.25 no.3
• /
• pp.43-62
• /
• 2019

At one time, the anomaly detection sector dominated the method of determining whether there was an abnormality based on the statistics derived from specific data. This methodology was possible because the dimension of the data was simple in the past, so the classical statistical method could work effectively. However, as the characteristics of data have changed complexly in the era of big data, it has become more difficult to accurately analyze and predict the data that occurs throughout the industry in the conventional way. Therefore, SVM and Decision Tree based supervised learning algorithms were used. However, there is peculiarity that supervised learning based model can only accurately predict the test data, when the number of classes is equal to the number of normal classes and most of the data generated in the industry has unbalanced data class. Therefore, the predicted results are not always valid when supervised learning model is applied. In order to overcome these drawbacks, many studies now use the unsupervised learning-based model that is not influenced by class distribution, such as autoencoder or generative adversarial networks. In this paper, we propose a method to detect anomalies using generative adversarial networks. AnoGAN, introduced in the study of Thomas et al (2017), is a classification model that performs abnormal detection of medical images. It was composed of a Convolution Neural Net and was used in the field of detection. On the other hand, sequencing data abnormality detection using generative adversarial network is a lack of research papers compared to image data. Of course, in Li et al (2018), a study by Li et al (LSTM), a type of recurrent neural network, has proposed a model to classify the abnormities of numerical sequence data, but it has not been used for categorical sequence data, as well as feature matching method applied by salans et al.(2016). So it suggests that there are a number of studies to be tried on in the ideal classification of sequence data through a generative adversarial Network. In order to learn the sequence data, the structure of the generative adversarial networks is composed of LSTM, and the 2 stacked-LSTM of the generator is composed of 32-dim hidden unit layers and 64-dim hidden unit layers. The LSTM of the discriminator consists of 64-dim hidden unit layer were used. In the process of deriving abnormal scores from existing paper of Anomaly Detection for Sequence data, entropy values of probability of actual data are used in the process of deriving abnormal scores. but in this paper, as mentioned earlier, abnormal scores have been derived by using feature matching techniques. In addition, the process of optimizing latent variables was designed with LSTM to improve model performance. The modified form of generative adversarial model was more accurate in all experiments than the autoencoder in terms of precision and was approximately 7% higher in accuracy. In terms of Robustness, Generative adversarial networks also performed better than autoencoder. Because generative adversarial networks can learn data distribution from real categorical sequence data, Unaffected by a single normal data. But autoencoder is not. Result of Robustness test showed that he accuracy of the autocoder was 92%, the accuracy of the hostile neural network was 96%, and in terms of sensitivity, the autocoder was 40% and the hostile neural network was 51%. In this paper, experiments have also been conducted to show how much performance changes due to differences in the optimization structure of potential variables. As a result, the level of 1% was improved in terms of sensitivity. These results suggest that it presented a new perspective on optimizing latent variable that were relatively insignificant.

• Journal of the Korea Institute of Information Security & Cryptology
• /
• v.18 no.5
• /
• pp.3-16
• /
• 2008

A Wireless Sensor Network (WSN for short) is a wireless network consisting of distributed small devices which are called sensor nodes or motes. Recently, there has been an extensive research on WSN and also on its security. For secure storage and secure transmission of the sensed information, sensor nodes should be equipped with cryptographic algorithms. Moreover, these algorithms should be efficiently implemented since sensor nodes are highly resource-constrained devices. There are already some existing algorithms applicable to sensor nodes, including public key ciphers such as TinyECC and standard block ciphers such as AES. Stream ciphers, however, are still to be analyzed, since they were only recently standardized in the eSTREAM project. In this paper, we implement over the MicaZ platform nine software-based stream ciphers out of the ten in the second and final phases of the eSTREAM project, and we evaluate their performance. Especially, we apply several optimization techniques to six ciphers including SOSEMANUK, Salsa20 and Rabbit, which have survived after the final phase of the eSTREAM project. We also present the implementation results of hardware-oriented stream ciphers and AES-CFB fur reference. According to our experiment, the encryption speeds of these software-based stream ciphers are in the range of 31-406Kbps, thus most of these ciphers are fairly acceptable fur sensor nodes. In particular, the survivors, SOSEMANUK, Salsa20 and Rabbit, show the throughputs of 406Kbps, 176Kbps and 121Kbps using 70KB, 14KB and 22KB of ROM and 2811B, 799B and 755B of RAM, respectively. From the viewpoint of encryption speed, the performances of these ciphers are much better than that of the software-based AES, which shows the speed of 106Kbps.

• Journal of the Korean Electrochemical Society
• /
• v.26 no.1
• /
• pp.1-10
• /
• 2023

The cathode, which is one of the four major components of a lithium secondary battery, is an important component responsible for the energy density of the battery. The mixing process of active material, conductive material, and polymer binder is very essential in the commonly used wet manufacturing process of the cathode. However, in the case of mixing conditions of the cathode, since there is no systematic method, in most cases, differences in performance occur depending on the manufacturer. Therefore, LiMn₂O₄ (LMO) cathodes were prepared using a commonly used THINKY mixer and homogenizer to optimize the mixing method in the cathode slurry preparation step, and their characteristics were compared. Each mixing condition was performed at 2000 RPM and 7 min, and to determine only the difference in the mixing method during the manufacture of the cathode other experiment conditions (mixing time, material input order, etc.) were kept constant. Among the manufactured THINKY mixer LMO (TLMO) and homogenizer LMO (HLMO), HLMO has more uniform particle dispersion than TLMO, and thus shows higher adhesive strength. Also, the result of the electrochemical evaluation reveals that HLMO cathode showed improved performance with a more stable life cycle compared to TLMO. The initial discharge capacity retention rate of HLMO at 69 cycles was 88%, which is about 4.4 times higher than that of TLMO, and in the case of rate capability, HLMO exhibited a better capacity retention even at high C-rates of 10, 15, and 20 C and the capacity recovery at 1 C was higher than that of TLMO. It's postulated that the use of a homogenizer improves the characteristics of the slurry containing the active material, the conductive material, and the polymer binder creating an electrically conductive network formed by uniformly dispersing the conductive material suppressing its strong electrostatic properties thus avoiding aggregation. As a result, surface contact between the active material and the conductive material increases, electrons move more smoothly, changes in lattice volume during charging and discharging are more reversible and contact resistance between the active material and the conductive material is suppressed.

• Journal of Intelligence and Information Systems
• /
• v.23 no.2
• /
• pp.71-88
• /
• 2017

Language models were originally developed for speech recognition and language processing. Using a set of example sentences, a language model predicts the next word or character based on sequential input data. N-gram models have been widely used but this model cannot model the correlation between the input units efficiently since it is a probabilistic model which are based on the frequency of each unit in the training set. Recently, as the deep learning algorithm has been developed, a recurrent neural network (RNN) model and a long short-term memory (LSTM) model have been widely used for the neural language model (Ahn, 2016; Kim et al., 2016; Lee et al., 2016). These models can reflect dependency between the objects that are entered sequentially into the model (Gers and Schmidhuber, 2001; Mikolov et al., 2010; Sundermeyer et al., 2012). In order to learning the neural language model, texts need to be decomposed into words or morphemes. Since, however, a training set of sentences includes a huge number of words or morphemes in general, the size of dictionary is very large and so it increases model complexity. In addition, word-level or morpheme-level models are able to generate vocabularies only which are contained in the training set. Furthermore, with highly morphological languages such as Turkish, Hungarian, Russian, Finnish or Korean, morpheme analyzers have more chance to cause errors in decomposition process (Lankinen et al., 2016). Therefore, this paper proposes a phoneme-level language model for Korean language based on LSTM models. A phoneme such as a vowel or a consonant is the smallest unit that comprises Korean texts. We construct the language model using three or four LSTM layers. Each model was trained using Stochastic Gradient Algorithm and more advanced optimization algorithms such as Adagrad, RMSprop, Adadelta, Adam, Adamax, and Nadam. Simulation study was done with Old Testament texts using a deep learning package Keras based the Theano. After pre-processing the texts, the dataset included 74 of unique characters including vowels, consonants, and punctuation marks. Then we constructed an input vector with 20 consecutive characters and an output with a following 21st character. Finally, total 1,023,411 sets of input-output vectors were included in the dataset and we divided them into training, validation, testsets with proportion 70:15:15. All the simulation were conducted on a system equipped with an Intel Xeon CPU (16 cores) and a NVIDIA GeForce GTX 1080 GPU. We compared the loss function evaluated for the validation set, the perplexity evaluated for the test set, and the time to be taken for training each model. As a result, all the optimization algorithms but the stochastic gradient algorithm showed similar validation loss and perplexity, which are clearly superior to those of the stochastic gradient algorithm. The stochastic gradient algorithm took the longest time to be trained for both 3- and 4-LSTM models. On average, the 4-LSTM layer model took 69% longer training time than the 3-LSTM layer model. However, the validation loss and perplexity were not improved significantly or became even worse for specific conditions. On the other hand, when comparing the automatically generated sentences, the 4-LSTM layer model tended to generate the sentences which are closer to the natural language than the 3-LSTM model. Although there were slight differences in the completeness of the generated sentences between the models, the sentence generation performance was quite satisfactory in any simulation conditions: they generated only legitimate Korean letters and the use of postposition and the conjugation of verbs were almost perfect in the sense of grammar. The results of this study are expected to be widely used for the processing of Korean language in the field of language processing and speech recognition, which are the basis of artificial intelligence systems.

• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 1993.06a
• /
• pp.1041-1043
• /
• 1993

Fuzzy logic based Control Theory has gained much interest in the industrial world, thanks to its ability to formalize and solve in a very natural way many problems that are very difficult to quantify at an analytical level. This paper shows a solution for treating membership function inside hardware circuits. The proposed hardware structure optimizes the memoried size by using particular form of the vectorial representation. The process of memorizing fuzzy sets, i.e. their membership function, has always been one of the more problematic issues for the hardware implementation, due to the quite large memory space that is needed. To simplify such an implementation, it is commonly [1,2,8,9,10,11] used to limit the membership functions either to those having triangular or trapezoidal shape, or pre-definite shape. These kinds of functions are able to cover a large spectrum of applications with a limited usage of memory, since they can be memorized by specifying very few parameters ( ight, base, critical points, etc.). This however results in a loss of computational power due to computation on the medium points. A solution to this problem is obtained by discretizing the universe of discourse U, i.e. by fixing a finite number of points and memorizing the value of the membership functions on such points [3,10,14,15]. Such a solution provides a satisfying computational speed, a very high precision of definitions and gives the users the opportunity to choose membership functions of any shape. However, a significant memory waste can as well be registered. It is indeed possible that for each of the given fuzzy sets many elements of the universe of discourse have a membership value equal to zero. It has also been noticed that almost in all cases common points among fuzzy sets, i.e. points with non null membership values are very few. More specifically, in many applications, for each element u of U, there exists at most three fuzzy sets for which the membership value is ot null [3,5,6,7,12,13]. Our proposal is based on such hypotheses. Moreover, we use a technique that even though it does not restrict the shapes of membership functions, it reduces strongly the computational time for the membership values and optimizes the function memorization. In figure 1 it is represented a term set whose characteristics are common for fuzzy controllers and to which we will refer in the following. The above term set has a universe of discourse with 128 elements (so to have a good resolution), 8 fuzzy sets that describe the term set, 32 levels of discretization for the membership values. Clearly, the number of bits necessary for the given specifications are 5 for 32 truth levels, 3 for 8 membership functions and 7 for 128 levels of resolution. The memory depth is given by the dimension of the universe of the discourse (128 in our case) and it will be represented by the memory rows. The length of a world of memory is defined by: Length = nem (dm(m)＋dm(fm) Where: fm is the maximum number of non null values in every element of the universe of the discourse, dm(m) is the dimension of the values of the membership function m, dm(fm) is the dimension of the word to represent the index of the highest membership function. In our case then Length=24. The memory dimension is therefore 128*24 bits. If we had chosen to memorize all values of the membership functions we would have needed to memorize on each memory row the membership value of each element. Fuzzy sets word dimension is 8*5 bits. Therefore, the dimension of the memory would have been 128*40 bits. Coherently with our hypothesis, in fig. 1 each element of universe of the discourse has a non null membership value on at most three fuzzy sets. Focusing on the elements 32,64,96 of the universe of discourse, they will be memorized as follows: The computation of the rule weights is done by comparing those bits that represent the index of the membership function, with the word of the program memor . The output bus of the Program Memory (μCOD), is given as input a comparator (Combinatory Net). If the index is equal to the bus value then one of the non null weight derives from the rule and it is produced as output, otherwise the output is zero (fig. 2). It is clear, that the memory dimension of the antecedent is in this way reduced since only non null values are memorized. Moreover, the time performance of the system is equivalent to the performance of a system using vectorial memorization of all weights. The dimensioning of the word is influenced by some parameters of the input variable. The most important parameter is the maximum number membership functions (nfm) having a non null value in each element of the universe of discourse. From our study in the field of fuzzy system, we see that typically nfm 3 and there are at most 16 membership function. At any rate, such a value can be increased up to the physical dimensional limit of the antecedent memory. A less important role n the optimization process of the word dimension is played by the number of membership functions defined for each linguistic term. The table below shows the request word dimension as a function of such parameters and compares our proposed method with the method of vectorial memorization[10]. Summing up, the characteristics of our method are: Users are not restricted to membership functions with specific shapes. The number of the fuzzy sets and the resolution of the vertical axis have a very small influence in increasing memory space. Weight computations are done by combinatorial network and therefore the time performance of the system is equivalent to the one of the vectorial method. The number of non null membership values on any element of the universe of discourse is limited. Such a constraint is usually non very restrictive since many controllers obtain a good precision with only three non null weights. The method here briefly described has been adopted by our group in the design of an optimized version of the coprocessor described in [10].

• Journal of Internet Computing and Services
• /
• v.14 no.6
• /
• pp.71-84
• /
• 2013

Log data, which record the multitude of information created when operating computer systems, are utilized in many processes, from carrying out computer system inspection and process optimization to providing customized user optimization. In this paper, we propose a MongoDB-based unstructured log processing system in a cloud environment for processing the massive amount of log data of banks. Most of the log data generated during banking operations come from handling a client's business. Therefore, in order to gather, store, categorize, and analyze the log data generated while processing the client's business, a separate log data processing system needs to be established. However, the realization of flexible storage expansion functions for processing a massive amount of unstructured log data and executing a considerable number of functions to categorize and analyze the stored unstructured log data is difficult in existing computer environments. Thus, in this study, we use cloud computing technology to realize a cloud-based log data processing system for processing unstructured log data that are difficult to process using the existing computing infrastructure's analysis tools and management system. The proposed system uses the IaaS (Infrastructure as a Service) cloud environment to provide a flexible expansion of computing resources and includes the ability to flexibly expand resources such as storage space and memory under conditions such as extended storage or rapid increase in log data. Moreover, to overcome the processing limits of the existing analysis tool when a real-time analysis of the aggregated unstructured log data is required, the proposed system includes a Hadoop-based analysis module for quick and reliable parallel-distributed processing of the massive amount of log data. Furthermore, because the HDFS (Hadoop Distributed File System) stores data by generating copies of the block units of the aggregated log data, the proposed system offers automatic restore functions for the system to continually operate after it recovers from a malfunction. Finally, by establishing a distributed database using the NoSQL-based Mongo DB, the proposed system provides methods of effectively processing unstructured log data. Relational databases such as the MySQL databases have complex schemas that are inappropriate for processing unstructured log data. Further, strict schemas like those of relational databases cannot expand nodes in the case wherein the stored data are distributed to various nodes when the amount of data rapidly increases. NoSQL does not provide the complex computations that relational databases may provide but can easily expand the database through node dispersion when the amount of data increases rapidly; it is a non-relational database with an appropriate structure for processing unstructured data. The data models of the NoSQL are usually classified as Key-Value, column-oriented, and document-oriented types. Of these, the representative document-oriented data model, MongoDB, which has a free schema structure, is used in the proposed system. MongoDB is introduced to the proposed system because it makes it easy to process unstructured log data through a flexible schema structure, facilitates flexible node expansion when the amount of data is rapidly increasing, and provides an Auto-Sharding function that automatically expands storage. The proposed system is composed of a log collector module, a log graph generator module, a MongoDB module, a Hadoop-based analysis module, and a MySQL module. When the log data generated over the entire client business process of each bank are sent to the cloud server, the log collector module collects and classifies data according to the type of log data and distributes it to the MongoDB module and the MySQL module. The log graph generator module generates the results of the log analysis of the MongoDB module, Hadoop-based analysis module, and the MySQL module per analysis time and type of the aggregated log data, and provides them to the user through a web interface. Log data that require a real-time log data analysis are stored in the MySQL module and provided real-time by the log graph generator module. The aggregated log data per unit time are stored in the MongoDB module and plotted in a graph according to the user's various analysis conditions. The aggregated log data in the MongoDB module are parallel-distributed and processed by the Hadoop-based analysis module. A comparative evaluation is carried out against a log data processing system that uses only MySQL for inserting log data and estimating query performance; this evaluation proves the proposed system's superiority. Moreover, an optimal chunk size is confirmed through the log data insert performance evaluation of MongoDB for various chunk sizes.

• Journal of Intelligence and Information Systems
• /
• v.23 no.2
• /
• pp.107-122
• /
• 2017

Volatility in the stock market returns is a measure of investment risk. It plays a central role in portfolio optimization, asset pricing and risk management as well as most theoretical financial models. Engle(1982) presented a pioneering paper on the stock market volatility that explains the time-variant characteristics embedded in the stock market return volatility. His model, Autoregressive Conditional Heteroscedasticity (ARCH), was generalized by Bollerslev(1986) as GARCH models. Empirical studies have shown that GARCH models describes well the fat-tailed return distributions and volatility clustering phenomenon appearing in stock prices. The parameters of the GARCH models are generally estimated by the maximum likelihood estimation (MLE) based on the standard normal density. But, since 1987 Black Monday, the stock market prices have become very complex and shown a lot of noisy terms. Recent studies start to apply artificial intelligent approach in estimating the GARCH parameters as a substitute for the MLE. The paper presents SVR-based GARCH process and compares with MLE-based GARCH process to estimate the parameters of GARCH models which are known to well forecast stock market volatility. Kernel functions used in SVR estimation process are linear, polynomial and radial. We analyzed the suggested models with KOSPI 200 Index. This index is constituted by 200 blue chip stocks listed in the Korea Exchange. We sampled KOSPI 200 daily closing values from 2010 to 2015. Sample observations are 1487 days. We used 1187 days to train the suggested GARCH models and the remaining 300 days were used as testing data. First, symmetric and asymmetric GARCH models are estimated by MLE. We forecasted KOSPI 200 Index return volatility and the statistical metric MSE shows better results for the asymmetric GARCH models such as E-GARCH or GJR-GARCH. This is consistent with the documented non-normal return distribution characteristics with fat-tail and leptokurtosis. Compared with MLE estimation process, SVR-based GARCH models outperform the MLE methodology in KOSPI 200 Index return volatility forecasting. Polynomial kernel function shows exceptionally lower forecasting accuracy. We suggested Intelligent Volatility Trading System (IVTS) that utilizes the forecasted volatility results. IVTS entry rules are as follows. If forecasted tomorrow volatility will increase then buy volatility today. If forecasted tomorrow volatility will decrease then sell volatility today. If forecasted volatility direction does not change we hold the existing buy or sell positions. IVTS is assumed to buy and sell historical volatility values. This is somewhat unreal because we cannot trade historical volatility values themselves. But our simulation results are meaningful since the Korea Exchange introduced volatility futures contract that traders can trade since November 2014. The trading systems with SVR-based GARCH models show higher returns than MLE-based GARCH in the testing period. And trading profitable percentages of MLE-based GARCH IVTS models range from 47.5% to 50.0%, trading profitable percentages of SVR-based GARCH IVTS models range from 51.8% to 59.7%. MLE-based symmetric S-GARCH shows +150.2% return and SVR-based symmetric S-GARCH shows +526.4% return. MLE-based asymmetric E-GARCH shows -72% return and SVR-based asymmetric E-GARCH shows +245.6% return. MLE-based asymmetric GJR-GARCH shows -98.7% return and SVR-based asymmetric GJR-GARCH shows +126.3% return. Linear kernel function shows higher trading returns than radial kernel function. Best performance of SVR-based IVTS is +526.4% and that of MLE-based IVTS is +150.2%. SVR-based GARCH IVTS shows higher trading frequency. This study has some limitations. Our models are solely based on SVR. Other artificial intelligence models are needed to search for better performance. We do not consider costs incurred in the trading process including brokerage commissions and slippage costs. IVTS trading performance is unreal since we use historical volatility values as trading objects. The exact forecasting of stock market volatility is essential in the real trading as well as asset pricing models. Further studies on other machine learning-based GARCH models can give better information for the stock market investors.