• International Journal of Internet, Broadcasting and Communication
• /
• v.16 no.1
• /
• pp.63-71
• /
• 2024

Deep reinforcement learning (RL) significantly influences autonomous vehicle development by optimizing decision-making and adaptation to complex driving environments through simulation-based training. In deep RL, an activation function is used, and various activation functions have been proposed, but their performance varies greatly depending on the application environment. Therefore, finding the optimal activation function according to the environment is important for effective learning. In this paper, we analyzed nine commonly used activation functions for RL to compare and evaluate which activation function is most effective when using deep RL for autonomous vehicles to learn highway merging. To do this, we built a performance evaluation environment and compared the average reward of each activation function. The results showed that the highest reward was achieved using Mish, and the lowest using SELU. The difference in reward between the two activation functions was 10.3%.

• Journal of Internet Computing and Services
• /
• v.22 no.4
• /
• pp.13-28
• /
• 2021

Reinforcement Learning (RL) is a machine learning algorithm that repeat the closed-loop process that agents perform actions specified by the policy, the action is evaluated with a reward function, and the policy gets updated accordingly. The key benefit of RL is the ability to optimze the policy with action evaluation. Hence, it can effectively be applied to developing advanced intelligent systems and autonomous systems. Conventional RL incoporates a single policy, a reward function, and relatively simple policy update, and hence its utilization was limited. In this paper, we propose an extended RL model that considers multiple instances of RL elements. We define a formal model of the key elements and their computing model of the extended RL. Then, we propose design methods for applying to system development. As a case stud of applying the proposed formal model and the design methods, we present the design and implementation of an advanced car navigator system that guides multiple cars to reaching their destinations efficiently.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.14 no.5
• /
• pp.131-135
• /
• 2014

The purpose of this study was to show the detection of the head movement within relatively the posture function. An analysis of the posture function was inquired a displacement that the ranges of stance direction showed generally a variation across all condition through the head movement. CNS condition (C_RL-MIN-AVG) was verified slightly greater variation at $0.226{\pm}0.04$ units. Somatosensory condition (So_$RL-_{MIN-AVG}$) was verified slightly greater variation at $0.939{\pm}0.46$ units. Vestibular condition (Ve_$RL-_{MIN-AVG}$) was verified slightly greater variation at $4.009{\pm}1.05$ units. Vision condition (Vi_$RL-_{MIN-AVG}$) was verified greater variation at $8.336{\pm}4.05$ units. When the movement head of vision characteristic function was presented a diminutive variance. On the CNS characteristic condition of the movement head function was presented a diminutive variance.

• Applied Chemistry for Engineering
• /
• v.29 no.6
• /
• pp.753-758
• /
• 2018

Surface activities including CMC, surface tension, foaming power etc. of the rhamnolipid (RL), a biosurfactant fermented by microorganism were investigated. Both the RL and liquid detergent contained RL showed a good surface activity. For the detergency test, the liquid detergent with RL showed a slightly better performance than that of using an fatty acid. However for the foaming profile, the liquid detergent with RL exhibited a lot of foam volume and the suppression of the rinsing and foaming during rinsing cycles with the RL was not seen. Therefore it can be concluded that RL can be used as a main surfactant or co-surfactant in liquid detergents without functions of foaming suppression or rinse-aid.

• 제어로봇시스템학회:학술대회논문집
• /
• 2000.10a
• /
• pp.489-489
• /
• 2000

Recently, Reinforcement Learning(RL) methods have been used far teaming problems in Partially Observable Markov Decision Process(POMDP) environments. Conventional RL-methods, however, have limited applicability to POMDP To overcome the partial observability, several algorithms were proposed [5], [7]. The aim of this paper is to extend our previous algorithm for POMDP, called Labeling Q-learning(LQ-learning), which reinforces incomplete information of perception with labeling. Namely, in the LQ-learning, the agent percepts the current states by pair of observation and its label, and the agent can distinguish states, which look as same, more exactly. Labeling is carried out by a hash-like function, which we call Labeling Function(LF). Numerous labeling functions can be considered, but in this paper, we will introduce several labeling functions based on only 2 or 3 immediate past sequential observations. We introduce the basic idea of LQ-learning briefly, apply it to maze problems, simple POMDP environments, and show its availability with empirical results, look better than conventional RL algorithms.

• Journal of the Korean Society of Fisheries and Ocean Technology
• /
• v.39 no.2
• /
• pp.143-151
• /
• 2003

A series of fishing experiments was carried out in the waters off the east coast of Korea from January, 2002 to March, 2003, using gill nets of different mesh sizes (m = 180, 210, 240, 270, and 300 ㎜) to determine the size selectivity of gill net for male snow crab Chionoecetes opilio. The maximum carapace length (RL) of each male snow crab caught in the fishing experiment was measured. The master curve of mesh selectivity was estimated by applying the extended Kitahara's method. Two kinds of functional models, quadratic function and cubic function were used to express logarithmic selectivity curve and were fitted to the data using the method of least squares. The obtained results were summarized as follows; 1. The cubic function of asymmetry was chosen to determine the selectivity curve of gill net for male snow crab from the model deviance comparison. 2. The result of size selectivity showed that the catch number of small male crab was getting decreased according to the increase of mesh size. 3. The optimum value (RL/m) was 0.549 and the RL/m was estimated to be 0.281, 0.296, and 0.356 when the retention probability were 0.2, 0.25 and 0.5, respectively.

• Korean Journal of Animal Reproduction
• /
• v.23 no.2
• /
• pp.95-103
• /
• 1999

This experiment was conducted to investigate how the lactation regulation such as restricted-lactation and early weaning during the suckling period influences on ovarian functions and change in serum levels of progesterone in primiparous rats. All the rats were raised in the individual cage from a few days before parturition through the suckling period. The normal lactation(NL) groups were controled 8 pups. The restricted-lactation(RL) and weaned(W) groups were subdivided into 5 subgroups as RL0, RL5, RL10, RL15 and RL20 as well as W0, W5, W10, W15, and W20 according to the day of onset of suckling. The number of pups were regulated from 8 to 4 on experimental strating day in RL gropus, and also perfectly weaned on the each on-set day in W groups. The results obtained were summarized as follows: 1. During the whole suckling period of 25 days the pups in RL group grew significantly(P＜005) faster than those in normal-lactation(NL) group. The pups in earlier RL group grew significantly(P＜0.05) faster than those in later RL rats, and there was no found any significant difference in body weight of pups between RL20 and NL group. The gestation period and litter size were found to be 21.53$\pm$0.04 days and 13.75$\pm$0.07, respectively. 2. The estrous cycle was not expressed in the NL group through the whole suckling period. An irregular estrous cycle was found around day 20 in RL0 group, and the regular estrous cycles were exhibited continuously from day 10 in the day 0 weaned rats. 3. In the rats of NL group the serum progesterone concentration increased from 33.16$\pm$2.64ng/$m\ell$ on day 0 to 122.5$\pm$53.68 ng/$m\ell$ on day 10, and then decreased slightly to 97.30$\pm$3.21 ng/$m\ell$ on day 20, but then decreased abruptly. However, the serum level of progesterone decreased greatly(P＜0.05) in 5 to 10 days following suckling restriction in the rats from which suckling began to be under restriction on day 0 or day 15. In the early weaning group the significant ( (P＜0.05) decrease in progesterone concentration was found similarly in 48 hours following weaning in all the rats weaned on day 0 through day 20. It was suggested that lactation stimulation is a very pivotal on the function of ovary.

• Corrosion Science and Technology
• /
• v.4 no.1
• /
• pp.15-18
• /
• 2005

Hydroxyapatite(HAp) was synthesized using the waste sludge from semiconductor process and used as an anticorrosive pigment. The water absorption of coating pigmented with anticorrosive pigment and the corrosion at interface between coating and substrate were monitored using AC impedance techniques. The anticorrosive performance of HAp was compared with those of red lead(RL) and zinc potassium chromate(ZPC), which have been known as representative anticorrosive pigments. The amount of absorbed water in ZPC- and HAp- pigmented coatings was much higher compared to that in RL-pigmented and unpigmented film. However, it seems that the water absorbed into HAp- or ZPC-pigmented film is beneficial to anticorrosive function. The anticorrosive performance of HAp is superior or at least comparable to those of ZPC and RL. The excellent anticorrosive properties of HAp can be explained by its passivating ability, caused by the reaction of the soluble component of HAp with Fe to form iron phosphate in the presence of water.

• The Korean Journal of Food And Nutrition
• /
• v.29 no.3
• /
• pp.335-340
• /
• 2016

The objective of the present investigation was to obtain vitamin, mineral, flavonoid, and polyphenol profiles of Rudbeckia laciniata (RL), and to examine the effects of extract of RL (RLE) on various physiological activities of HaCaT keratinocyte for the utilization of RL as natural raw materials to develop functional food. To accomplish this purpose, we checked the contents of the general nutrients of RL. The contents of vitamin A, vitamin $B_1$ and vitamin $B_2$ were $7.49{\mu}g/g$, $51.96{\mu}g/g$, and $132{\mu}g/g$ respectively, while vitamin C and vitamin $D_3$ were not detected. The contents of mineral such as Ca, K and Fe were 2.01 mg/g, 6.06 mg/g and 0.03 mg/g respectively. Total flavonoid contents of RLE were 0.25 mg/g, and total polyphenol were estimated as 1.43 mg/g. Because RL contains high levels of vitamin A which is associated with skin aging, we investigated the effect of RLE on physiological function of keratinocytes with respect to skin aging. We found that RLE significantly increased the growth rate of HaCaT cells and reduced ultraviolet radiation B (UVB)-induced cellular toxicity. Also, the extract of Rudbeckia laciniata attenuated the UVB-induced reactive oxygen species (ROS) generation in a dose-dependent manner in HaCaT cells. In addition, treatment with the extract dose-dependently increased migration activity of HaCaT cells. Thus, these findings indicated that RLE could regulate the physiological activity of keratinocytes, and may be used to develop functional foods.

• Sleep Medicine and Psychophysiology
• /
• v.22 no.1
• /
• pp.20-24
• /
• 2015

Objectives: Restless legs syndrome (RLS) is a common sleep disorder in adults with diabetes. This study investigated the frequency of RLS and clinical correlations in children and adolescents with type 1 diabetes. Methods: This study included 55 consecutive patients (21 males, age $12.6{\pm}3.4years$) with type I diabetes that were regularly treated at the Department of Pediatric Endocrinology. RLS was diagnosed by intensive interviews which also included the Epworth Sleepiness Scale (ESS) and International RLS Rating Scale (IRLSRS). Patients also received neurological examinations and laboratory tests for diabetes, iron metabolism and renal function. Results: Thirteen patients (23.6%, 6 males) were compatible for the diagnostic criteria of RLS. None of the RLS patients showed abnormal findings in neurological evaluations and 7 patients had familial history of RLS. Demographic and laboratory findings were not different between the patients with or without RLS. The RLS group showed significantly increased ESS and IRLSRS scores. Conclusion: RLS was prevalent in children and adolescents with type 1 diabetes. The association between RLS and diabetes-related laboratory findings requires further investigation.