• Journal of Life Science
• /
• v.7 no.3
• /
• pp.186-191
• /
• 1997

In order to select new anticandiosis agent-producing candidates, we used a modified selection method. Two active fractions designated as S-6279A) and (B) were isolated from the fermentation broth of Streptomycetes sp. S627 which was screened by methanol extraction, diaion HP-20 and silica gel column chromatography (chloroform-me-thanol, and benzene-ethyl acetate), and preparative thin layer chromatography.

• The Journal of Korea Robotics Society
• /
• v.14 no.3
• /
• pp.157-162
• /
• 2019

The video game Tetris is one of most popular game and it is well known that its game rule can be modelled as MDP (Markov Decision Process). This paper presents a DQN (Deep Q-Network) based game agent for Tetris game. To this end, the state is defined as the captured image of the Tetris game board and the reward is designed as a function of cleared lines by the game agent. The action is defined as left, right, rotate, drop, and their finite number of combinations. In addition to this, PER (Prioritized Experience Replay) is employed in order to enhance learning performance. To train the network more than 500000 episodes are used. The game agent employs the trained network to make a decision. The performance of the developed algorithm is validated via not only simulation but also real Tetris robot agent which is made of a camera, two Arduinos, 4 servo motors, and artificial fingers by 3D printing.

• International Journal of Internet, Broadcasting and Communication
• /
• v.14 no.2
• /
• pp.192-198
• /
• 2022

Companies are building and utilizing their own data analysis systems according to business characteristics in the distributed cloud. However, as businesses and data types become more complex and diverse, the demand for more efficient analytics has increased. In response to these demands, in this paper, we propose an unsupervised learning-based data analysis agent to which reinforcement learning is applied for effective data analysis. The proposal agent consists of reinforcement learning processing manager and unsupervised learning manager modules. These two modules configure an agent with k-means clustering on multiple nodes and then perform distributed training on multiple data sets. This enables data analysis in a relatively short time compared to conventional systems that perform analysis of large-scale data in one batch.

• Korean Journal of Child Studies
• /
• v.38 no.1
• /
• pp.205-215
• /
• 2017

Objective: The current study examined whether providing both an actor's eye gaze and emotional expressions can help 10-month-olds interpret a change in the actor's words as a signal to a change in the actor's goal object. Methods: Sixteen 10-month-olds participated in an experiment using the violation-of-expectation paradigm and were compared to 16 10-month-olds in a control condition. The infants in the experimental condition were familiarized to an event in which an actor looks at one of two novel objects, excitingly utters a sentence, "Wow, here's a modi!", and grasps the object. The procedure in the control condition was identical to that of the experimental condition except that the infants heard the sentence without any emotional excitement and the eye gaze of the agent was hidden by a visor. In the following test trial, the infants in both conditions heard the agent changing her word (from modi to papu) and watched her grasping either the same object as before (old-goal event) or the new object (new-goal event). Results: The infants in the experimental condition looked at the old-goal event longer than at the new-goal event, suggesting that they expected the agent to change her goal object when the actor changed her word. However, the infants in the control condition looked at the two events about equally. Conclusion: When both eye gaze and emotional cues were provided, 10-month-olds were able to exploit the agent's verbal information when reasoning about whether the agent would pursue the same goal object as before.

• Archives of Reconstructive Microsurgery
• /
• v.16 no.2
• /
• pp.63-67
• /
• 2007

This study was designed to introduce free toe tissue transfer using infiltration method of local anesthetic agent. Four toe pulp tissues were transferred to reconstruct finger tip defect in four patients who were not suitable for general anesthesia. Two flaps taken from the lateral side of the great toe was used for reconstruction of thumb defect and two flaps from the medial side of the second toe for resurfacing of the index and fifth finger. Flap sizes were various from $2.0{\times}2.0\;cm^2$ to $1.6{\times}4.0\;cm^2$. Anesthesia was induced by infiltration of 2% lidocaine hydrochloride (with 1:100,000 epinephrine) with dilution of normal saline in same volume unit, as like as in ordinary digital block. All vessels were anastomosed within 2 cm of distance from the proximal margin of the defect. Whole operative procedures were carried out by one team. All flaps were successfully taken without complication. The average operation time was 4 hours 10 minutes. The amount of anesthetic agent used in whole operative procedures was roughly 4 mL in the toe, 8 mL in the finger, and 12 mL totally. In conclusion, free toe tissue transfer using infiltration method of local anesthetic agent would be good strategy for finger tip reconstruction in the patient not suitable for general anesthesia.

• Journal of the Korean Institute of Telematics and Electronics S
• /
• v.36S no.7
• /
• pp.25-33
• /
• 1999

This paper proposes a handover algorithm for the mobile host in order to reduce the number of unnecessary retransmissions during handover. In our proposal, the home agent buffers user data after receiving the registration request message form the foreign agent and then transmits them again after a predefined time. We also analyze the packet disordering probability between two user data under an assumption that the home agent generates all user data. One of two user data is the one which has been generated just before the home agent receives the registration request message and the other is generated just after that. Simulation results show that our proposal provides efficient handover for a mobile host by reducing the packet disordering probalility.

• Applied Biological Chemistry
• /
• v.38 no.6
• /
• pp.522-527
• /
• 1995

To develop an antiviral agent for the rice black-streaked dwarf virus (RBSDV), a hammerhead type ribozyme, which has a potential target site on the genome segment 3, was designed. Oligonucleotides for the ribozyme and its substrate were synthesized, annealed, and cloned into a plasmid pBluescript II KS(+). Ribozyme and substrate RNAs were then synthesized by in vitro transcription with $T_3$ RNA polymerase, obtaining RNAs in expected size, 193 and 182 nucleotides, respectively. The substrate RNA was efficiently cleaved into two fragments when incubated with the ribozyme at $55^{\circ}C$, while the cleavage was not detected at $37^{\circ}C$. In addition, the segment 3 RNA of RBSDV was also cleaved into two fragments by the same ribozyme at $55^{\circ}C$. Taken together, our results demonstrated that the hammerhead ribozyme has an in vitro endonucleolytic activity and may be used as an antiviral agent in transgenic plants.

• Textile Coloration and Finishing
• /
• v.12 no.1
• /
• pp.44-51
• /
• 2000

The conventional one bath exhaust dyeing system of nylon]/cotton blends with acid dye/direct dye involves a lot of neutral salt which affect dyeability of acid dyes. Because of conventional one bath exhaust dyeing system of nylonl/cotton blends with acid dye/direct dye needs acid dyes adjusted at a neutral liquor, suitable acid dye is limited. To improve dyeing property of nylon/cotton blends, cotton component was pretreated with cationizing agent containing chlorohydrine group in aqueous solution of sodium hydroxide. This study was carried out to investigate dyeing possibilities of nylon/cationized cotton blends with acid dye/direct dye in a non-neutral salt dyebath by one bath two step method. The concentration of direct dye was 1.0%, 0.7%, 0.3%o.w.f. respectively at a non, 1%, 3%owf of reserving agent for being almost equal color strength between nylon and 3% cationized cotton when nylon/3% cationized cotton blends was dyed with acid dye/direct dye at concentration of acid dye(1%o.w.f.) by one bath two step method.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.14 no.9
• /
• pp.2143-2149
• /
• 2010

An agent based geographic routing protocol is proposed to improve the well-known geographic routing protocol-GPSR routing protocol. In the proposed scheme, the agent is selected by sink node which concern about the source node's position as well as agent candidate's state. So packets will first be forwarded to agent and next step is to be forwarded to their final goal- sink node from agent. During the next hop selection process, nodes select their neighbors by considering not only position but also their average available buffer size. This results in efficient selection of next hop node in congestion area, and then increases the successful packet delivery ratio. The simulation is conducted for two scenarios: general number of connections and large number of connections in our map. Results show that new method with agent achieves improved performance in successful packet delivery ratio when compares to GPSR without our scheme.

• The KIPS Transactions:PartC
• /
• v.11C no.1
• /
• pp.31-38
• /
• 2004

In mobile agent environments, migration of an agent occurs continuously due to the mobility of agents. So cascaded delegation can occur among places for delegating the privilege to execute the agent. Because the existing delegation scheme considers only the delegation between two places that participate in migration of an agent, it does not support secure cascaded delegation. In this paper, we propose a cascaded delegation scheme that provides agents with secure cascaded delegation in mobile agent environments. Depending on the trust-relationship among places, the proposed scheme achieves the goal by nesting each delegation token or by nesting only initial token signed by creator of the agent within the signed part of the next immediate delegation token. And we prove that the proposed scheme is secure against the attack of replaying a message and the attack of substituting a delegation token.