• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.15 no.5
• /
• pp.1590-1609
• /
• 2021

Laughter is one of the most important nonverbal sound that human generates. It is a means for expressing his emotions. The acoustic and contextual features of this specific sound are different from those of speech and many difficulties arise during their modeling process. During this work, we propose an audio laughter generation system based on unsupervised generative models: the autoencoder (AE) and its variants. This procedure is the association of three main sub-process, (1) the analysis which consist of extracting the log magnitude spectrogram from the laughter database, (2) the generative models training, (3) the synthesis stage which incorporate the involvement of an intermediate mechanism: the vocoder. To improve the synthesis quality, we suggest two hybrid models (LSTM-VAE, GRU-VAE and CNN-VAE) that combine the representation learning capacity of variational autoencoder (VAE) with the temporal modelling ability of a long short-term memory RNN (LSTM) and the CNN ability to learn invariant features. To figure out the performance of our proposed audio laughter generation process, objective evaluation (RMSE) and a perceptual audio quality test (listening test) were conducted. According to these evaluation metrics, we can show that the GRU-VAE outperforms the other VAE models.

• Journal of Korea Game Society
• /
• v.21 no.6
• /
• pp.87-96
• /
• 2021

As the industry develops, the technology used for games is also being advanced. In particular, AI technology is used to game automation and intelligence. These game player patterns are widely used in online games such as player matchmaking, generation of friendly or hostile NPCs, and balancing of game worlds. This study proposes a model generation method for game players. For model generation, attributes such as hunting, collection, movement, combat, crisis management, production, and interaction were defined, and patterns were extracted and modeled using decision tree method. To evaluate the proposed method, we used the game log of a commercial game and confirmed the meaningful results.

• Proceedings of the Korea Contents Association Conference
• /
• 2007.11a
• /
• pp.951-953
• /
• 2007

Generation of Corporate risk Contents using Financial Data The purpose of this paper is to capture risk profiles of smaller-sized Korean firms vis-$\grave{a}$-vis larger-sized firms during the Asian financial crisis. For this purpose, risk profiles are provided by estimating expected default risks and by tracking how these have changed during this period with respect to their magnitude, volatility, and sensitivity measures. Methodology used in this study employs the Black-Scholes-Merton model for producing estimates of default risks. And the conventional trans-log function is utilized for obtaining sensitivity measures of the estimated default risks. According to empirical evidence obtained here, it is revealed that contractions of corporate loans associated with IMF austerity policy was the main factor responsible for the drastic change in the default risk profile of Korean firms after occurrence of the Asian financial crisis.

• The Journal of the Acoustical Society of Korea
• /
• v.21 no.4
• /
• pp.407-414
• /
• 2002

In this paper, we present a method that automatically generates concatenate morpheme based language models to improve the performance of Korean large vocabulary continuous speech recognition. The focus was brought into improvement against recognition errors of monosyllable morphemes that occupy 54% of the training text corpus and more frequently mis-recognized. Knowledge-based method using POS patterns has disadvantages such as the difficulty in making rules and producing many low frequency concatenate morphemes. Proposed method automatically selects morpheme-pairs from training text data based on measures such as frequency, mutual information, and unigram log likelihood. Experiment was performed using 7M-morpheme text corpus and 20K-morpheme lexicon. The frequency measure with constraint on the number of morphemes used for concatenation produces the best result of reducing monosyllables from 54% to 30%, bigram perplexity from 117.9 to 97.3. and MER from 21.3% to 17.6%.

• Journal of KIISE:Computing Practices and Letters
• /
• v.14 no.2
• /
• pp.236-240
• /
• 2008

People use diary or photograph for recall-ing their memory in order to satisfy their desires for recording their lives. If the experienced events are organized to a story, S/he can share her/his experience with others, and recall her/his significant events easily. In this paper, we propose a method that generates a story with Petri net and user contexts collected from mobile device. Here, we use Petri-net as a representation method that links human activities or experience causally. It is appropriate solution for modeling parallel events in real world, and for representing non-linear story line. In order to show the usefulness of the proposed method, we show an example of generating a story of user's experience with user contexts from mobile device and evaluate them.

• Journal of Food Hygiene and Safety
• /
• v.34 no.6
• /
• pp.583-590
• /
• 2019

As the consumption of fresh fruits and vegetables increases, food poisoning caused by foodborne pathogen contamination is not decreasing. To prevent the contamination of produce, a quick and easy, low-cost, environmentally-safe disinfection method that does not affect produce freshness or quality is needed. This study demonstrates a new-concept, circulating-water disinfection system that purifies water by using newly developed 'LED-PS (photosensitizer)-induced ROS generation unit'. Using various types of LED-PS induced ROS generation units, we investigated the conditions for reducing the density of various pathogenic bacteria by more than 3 log CFU / mL in 1 hour. The major operational factors affecting the density reduction of the LED-PS-induced ROS generation unit were analyzed. Depending on bacteria species, the density reduction rate was varied. The effect of the units on reducing the density of Bacillus cereus and Pectobacterium carotovorum subsp. carotovorum was high, but the effect on foodborne bacteria such as Escherichia coli was relatively low. In this circulating water disinfection system, the density reduction effect tended to increase as the flow rate increased and the initial bacterial density decreased. As the amount of PS absorbed beads increased, the density reduction effect increased exponentially in some bacteria. Model 3280, a double cylindrical unit connecting two single cylindrical units, could completely sterilize more than 3 log CFU/mL of B. cereus and P. carotovorum subsp. carotovorum in 30 minutes of LED irradiation.

• Geomechanics and Engineering
• /
• v.13 no.1
• /
• pp.99-117
• /
• 2017

Rock physics modeling of sandstone reservoir from gas fields of Krishna-Godavari basin represents the link between reservoir parameters and seismic properties. The rock physics diagnostic models such as contact cement, constant cement and friable sand are chosen to characterize reservoir sands of two wells in this basin. Cementation is affected by the grain sorting and cement coating on the surface of the grain. The models show that the reservoir sands in two wells under examination have varying cementation from 2 to more than 6%. Distinct and separate velocity-porosity and elastic moduli-porosity trends are observed for reservoir zones of two wells. A methodology is adopted for generation of Rock Physics Template (RPT) based on fluid replacement modeling for Raghavapuram Shale and Gollapalli Sandstones of Early Cretaceous. The ratio of P-wave velocity to S-wave velocity (Vp/Vs) and P-impedance template, generated for this above formations is able to detect shale, brine sand and gas sand with varying water saturation and porosity from wells in the Endamuru and Suryaraopeta gas fields having same shallow marine depositional characters. This RPT predicted detection of water and gas sands are matched well with conventional neutron-density cross plot analysis.

• The Korean Journal of Mycology
• /
• v.41 no.1
• /
• pp.28-32
• /
• 2013

New Shiitake (Lentinula edodes) strain "Chunbaegko" was bred by Di-mon method. Bed-log cultivation of "Chunbaegko" was performed. Fruit-body production of "Chunbaegko" was most at spring and autumn. Optimal temperature of fruit-body formation was $14{\sim}22^{\circ}C$ Baegwhako, the best quality fruit-body, is produced during spring. The diameter of pileus is ca. 52 mm. The total amount of fruit-body production during 4 years(one generation) was 140 $kg/m^3$ log.

• Journal of Mushroom
• /
• v.11 no.1
• /
• pp.9-14
• /
• 2013

New shiitake (Lentinula edodes) strain "soohyangko" was bred for bed-log cultivation. Fruit-body production of "soohyangko" was most at summer and autumn. The fruit-body is hemispherical shape, brown colored and diameter of pileus is ca. 56 mm. Optimum temperature of fruit-body formation was $18{\sim}28^{\circ}$, and the fruiting is concentrated. The total amount of fruit-body production during 4 years (one generation) was $140kg/m^3$ log.

• Journal of Internet Computing and Services
• /
• v.12 no.3
• /
• pp.111-117
• /
• 2011

In this paper, the BER(Bit Error Rate) performance of a DVB-T2(Second Generation Digital Terrestrial Television Broadcasting System) in MISO(Multiple Input Single Output) transmission mode is evaluated by the computer simulation. In the DVB-T2 receiver, an IDD(Iterative Demapping and Decoder) technique is employed that exchanges extrinsic information between the demapper and the LDPC decoder. Simulation results show that the IDD-based DVB-T2 receiver in MISO transmission mode provides 2dB gain at BER of $10^{-4}$ but suffer from the frequency offsets between transmit antennas.