• Title/Summary/Keyword: Adaptive loss function

Search Result 43, Processing Time 0.043 seconds

Vehicle detection and tracking algorithm based on improved feature extraction

  • Xiaole Ge;Feng Zhou;Shuaiting Chen;Gan Gao;Rugang Wang
    • KSII Transactions on Internet and Information Systems (TIIS)
    • /
    • v.18 no.9
    • /
    • pp.2642-2664
    • /
    • 2024
  • In the process of modern traffic management, information technology has become an important part of intelligent traffic governance. Real-time monitoring can accurately and effectively track and record vehicles, which is of great significance to modern urban traffic management. Existing tracking algorithms are affected by the environment, viewpoint, etc., and often have problems such as false detection, imprecise anchor boxes, and ID switch. Based on the YOLOv5 algorithm, we improve the loss function, propose a new feature extraction module to obtain the receptive field at different scales, and do adaptive fusion with the SGE attention mechanism, so that it can effectively suppress the noise information during feature extraction. The trained model improves the mAP value by 5.7% on the public dataset UA-DETRAC without increasing the amount of calculations. Meanwhile, for vehicle feature recognition, we adaptively adjust the network structure of the DeepSort tracking algorithm. Finally, we tested the tracking algorithm on the public dataset and in a realistic scenario. The results show that the improved algorithm has an increase in the values of MOTA and MT etc., which generally improves the reliability of vehicle tracking.

PRICE ESTIMATION VIA BAYESIAN FILTERING AND OPTIMAL BID-ASK PRICES FOR MARKET MAKERS

  • Hyungbin Park;Junsu Park
    • Journal of the Korean Mathematical Society
    • /
    • v.61 no.5
    • /
    • pp.875-898
    • /
    • 2024
  • This study estimates the true price of an asset and finds the optimal bid/ask prices for market makers. We provide a novel state-space model based on the exponential Ornstein-Uhlenbeck volatility and the Heston models with Gaussian noise, where the traded price and volume are available, but the true price is not observable. An objective of this study is to use Bayesian filtering to estimate the posterior distribution of the true price, given the traded price and volume. Because the posterior density is intractable, we employ the guided particle filtering algorithm, with which adaptive rejection metropolis sampling is used to generate samples from the density function of an unknown distribution. Given a simulated sample path, the posterior expectation of the true price outperforms the traded price in estimating the true price in terms of both the mean absolute error and root-mean-square error metrics. Another objective is to determine the optimal bid/ask prices for a market maker. The profit-and-loss of the market maker is the difference between the true price and its bid/ask prices multiplied by the traded volume or bid/ask size of the market maker. The market maker maximizes the expected utility of the PnL under the posterior distribution. We numerically calculate the optimal bid/ask prices using the Monte Carlo method, finding that its spread widens as the market maker becomes more risk-averse, and the bid/ask size and the level of uncertainty increase.

Robust Object Tracking based on Weight Control in Particle Swarm Optimization (파티클 스웜 최적화에서의 가중치 조절에 기반한 강인한 객체 추적 알고리즘)

  • Kang, Kyuchang;Bae, Changseok;Chung, Yuk Ying
    • The Journal of Korean Institute of Next Generation Computing
    • /
    • v.14 no.6
    • /
    • pp.15-29
    • /
    • 2018
  • This paper proposes an enhanced object tracking algorithm to compensate the lack of temporal information in existing particle swarm optimization based object trackers using the trajectory of the target object. The proposed scheme also enables the tracking and documentation of the location of an online updated set of distractions. Based on the trajectories information and the distraction set, a rule based approach with adaptive parameters is utilized for occlusion detection and determination of the target position. Compare to existing algorithms, the proposed approach provides more comprehensive use of available information and does not require manual adjustment of threshold values. Moreover, an effective weight adjustment function is proposed to alleviate the diversity loss and pre-mature convergence problem in particle swarm optimization. The proposed weight function ensures particles to search thoroughly in the frame before convergence to an optimum solution. In the existence of multiple objects with similar feature composition, this algorithm is tested to significantly reduce convergence to nearby distractions compared to the other existing swarm intelligence based object trackers.

Performance Evaluation of ResNet-based Pneumonia Detection Model with the Small Number of Layers Using Chest X-ray Images (흉부 X선 영상을 이용한 작은 층수 ResNet 기반 폐렴 진단 모델의 성능 평가)

  • Youngeun Choi;Seungwan Lee
    • Journal of radiological science and technology
    • /
    • v.46 no.4
    • /
    • pp.277-285
    • /
    • 2023
  • In this study, pneumonia identification networks with the small number of layers were constructed by using chest X-ray images. The networks had similar trainable-parameters, and the performance of the trained models was quantitatively evaluated with the modification of the network architectures. A total of 6 networks were constructed: convolutional neural network (CNN), VGGNet, GoogleNet, residual network with identity blocks, ResNet with bottleneck blocks and ResNet with identity and bottleneck blocks. Trainable parameters for the 6 networks were set in a range of 273,921-294,817 by adjusting the output channels of convolution layers. The network training was implemented with binary cross entropy (BCE) loss function, sigmoid activation function, adaptive moment estimation (Adam) optimizer and 100 epochs. The performance of the trained models was evaluated in terms of training time, accuracy, precision, recall, specificity and F1-score. The results showed that the trained models with the small number of layers precisely detect pneumonia from chest X-ray images. In particular, the overall quantitative performance of the trained models based on the ResNets was above 0.9, and the performance levels were similar or superior to those based on the CNN, VGGNet and GoogleNet. Also, the residual blocks affected the performance of the trained models based on the ResNets. Therefore, in this study, we demonstrated that the object detection networks with the small number of layers are suitable for detecting pneumonia using chest X-ray images. And, the trained models based on the ResNets can be optimized by applying appropriate residual-blocks.

Aortic Remodelling in Chronic Nicotine-Administered Rat

  • Zainalabidin, Satirah;Budin, Siti Balkis;Ramalingam, Anand;Lim, Yi Cheng
    • The Korean Journal of Physiology and Pharmacology
    • /
    • v.18 no.5
    • /
    • pp.411-418
    • /
    • 2014
  • Vascular remodelling is an adaptive mechanism, which counteracts pressure changes in blood circulation. Nicotine content in cigarette increases the risk of hypertension. The exact relationship between nicotine and vascular remodelling still remain unknown. Current study was aimed to determine the effect of clinically relevant dosage of nicotine (equivalent to light smoker) on aortic reactivity, oxidative stress markers and histomorphological changes. Twelve age-matched male Sprague-Dawley rats were randomly divided into two groups, i.e.: normal saline as control or 0.6 mg/kg nicotine for 28 days (i.p., n=6 per group). On day-29, the rats were sacrificed and the thoracic aorta was dissected immediately for further studies. Mean arterial pressure (MAP) and pulse pressure (PP) of nicotine-treated vs. control were significantly increased (p<0.05). Nicotine-treated group showed significant (p<0.05) increase tunica media thickness, and decrease in lumen diameter, suggesting vascular remodelling which lead to prior hypertension state. The phenylephrine (PE)-induced contractile response in nicotine group was significantly higher than control group ($ED_{50}=1.44{\times}10^5M$ vs. $4.9{\times}10^6M$) (p<0.05~0.001). However, nicotine-treated rat showed significantly lower endothelium-dependent relaxation response to acetylcholine (ACh) than in control group ($ED_{50}=6.17{\times}10^7M$ vs. $2.82{\times}10^7M$) (p<0.05), indicating loss of primary vascular function. Malondialdehyde (MDA), a lipid peroxidation marker was significantly higher in nicotine group. Superoxide dismutase (SOD) enzymatic activity and glutathione (GSH) were all reduced in nicotine group (p<0.05) vs. control, suggesting nicotine induces oxidative imbalance. In short, chronic nicotine administration impaired aortic reactivity, probably via redox imbalance and vascular remodelling mechanism.

A Study on Evaluation Method of AEB Test (AEB 시험평가 방법에 관한 연구)

  • Kim, BongJu;Lee, SeonBong
    • Journal of Auto-vehicle Safety Association
    • /
    • v.10 no.2
    • /
    • pp.20-28
    • /
    • 2018
  • Currently, sharp increase of car is on the rise as a serious social problem due to loss of lives from car accident and environmental pollution. There is a study on ITS (Intelligent Transportation System) to seek coping measures. As for the commercialization of ITS, we aim for occupancy of world market through ASV (Advanced Safety Vehicle) related system development and international standardization. However, the domestic environment is very insufficient. Core factor technologies of ITS are Adaptive Cruise Control, Lane Keeping Assist System, Forward Collision Warning System, AEB (Autonomous Emergency Braking) system etc. These technologies are applied to cars to support driving of a driver. AEB system is stop the car automatically based on the result decided by the relative speed and distance with obstacle detected through sensor attached on car rather than depending on the driver. The purpose of AEB system is to measure the distance and speed of car and to prevent accident. Thus, AEB will be a system useful for prevention of accident by decreasing car accident along with the development of automobile technology. This study suggests a scenario to suggest a test evaluation method that accords with domestic environment and active response of international standard regarding the test evaluation method of AEB. Also, by setting the goal with function for distance, it suggests theoretic model according to the result. And the study aims to verify the theoretic evaluation standard per proposed scenario using car which is installed with AEB device through field car driving test on test road. It will be useful to utilize the suggested scenario and theoretical model when conducting AEB test evaluation.

The Role of MicroRNAs in Regulatory T Cells and in the Immune Response

  • Ha, Tai-You
    • IMMUNE NETWORK
    • /
    • v.11 no.1
    • /
    • pp.11-41
    • /
    • 2011
  • The discovery of microRNA (miRNA) is one of the major scientific breakthroughs in recent years and has revolutionized current cell biology and medical science. miRNAs are small (19~25nt) noncoding RNA molecules that post-transcriptionally regulate gene expression by targeting the 3' untranslated region (3'UTR) of specific messenger RNAs (mRNAs) for degradation of translation repression. Genetic ablation of the miRNA machinery, as well as loss or degradation of certain individual miRNAs, severely compromises immune development and response, and can lead to immune disorders. Several sophisticated regulatory mechanisms are used to maintain immune homeostasis. Regulatory T (Treg) cells are essential for maintaining peripheral tolerance, preventing autoimmune diseases and limiting chronic inflammatory diseases. Recent publications have provided compelling evidence that miRNAs are highly expressed in Treg cells, that the expression of Foxp3 is controlled by miRNAs and that a range of miRNAs are involved in the regulation of immunity. A large number of studies have reported links between alterations of miRNA homeostasis and pathological conditions such as cancer, cardiovascular disease and diabetes, as well as psychiatric and neurological diseases. Although it is still unclear how miRNA controls Treg cell development and function, recent studies certainly indicate that this topic will be the subject of further research. The specific circulating miRNA species may also be useful for the diagnosis, classification, prognosis of diseases and prediction of the therapeutic response. An explosive literature has focussed on the role of miRNA. In this review, I briefly summarize the current studies about the role of miRNAs in Treg cells and in the regulation of the innate and adaptive immune response. I also review the explosive current studies about clinical application of miRNA.

Full mouth rehabilitation of edentulous patient with intellectual disability using implants and monolithic zirconia (1급 지적장애 환자에서 임플란트와 단일 구조 지르코니아를 이용한 완전구강회복 증례)

  • Jeong, Ki-Won;Kim, Sung-Hun;Han, Jung-Suk;Yeo, In-Sung;Yoon, Hyung-In
    • The Journal of Korean Academy of Prosthodontics
    • /
    • v.55 no.2
    • /
    • pp.156-163
    • /
    • 2017
  • People with class I intellectual disability need lifelong assistance and protection from their surroundings due to impaired adaptive functioning. They have poor oral health and show higher prevalence of dental caries, periapical inflammation and tooth loss that require proper prosthetic restoration. Because removable prostheses for intellectually disabled patients often lack stability, retention, and maintenance, fixed prostheses are essential and the only available option is dental implants. In this case, a 45 year-old male patient with class I intellectual disability had poor oral hygiene with most of his teeth missing and visited the clinic to recover his masticatory function. Due to such systemic conditions, the definitive restoration of choice was the implant-supported fixed dental prosthesis made of biocompatible and highly strong monolithic zirconia. In consequence of the treatment process, the patient was able to improve his oral environment aesthetically and functionally.

Ischemic Time Associated with Activation of Rejection-Related Immune Responses (허혈 시간과 거부반응 관련 면역반응)

  • Nam, Hyun-Suk;Choi, Jin-Yeung;Kim, Yoon-Tai;Kang, Kyung-Sun;Kwon, Hyuk-Moo;Hong, Chong-Hae;Kim, Doo;Han, Tae-Wook;Moon, Tae-Young;Kim, Jee-Hee;Cho, Byung-Ryul;Woo, Heung-Myong
    • Journal of Veterinary Clinics
    • /
    • v.26 no.2
    • /
    • pp.138-143
    • /
    • 2009
  • Ischemia/reperfusion injury(I/RI) is the major cause of acute renal failure and delayed graft function(DGF) unavoidable in renal transplantation. Enormous studies on ischemia damage playing a role in activating graft rejection factors, such as T cells or macrophages, are being reported. Present study was performed to determine whether ischemia time would play an important role in activating rejection-related factors or not in rat models of I/RI. Male Sprague-Dawley rats were submitted to 30, 45, and 60 minutes of warm renal ischemia with nephrectomy or control animals underwent sham operation(unilateral nephrectomy). Renal function and survival rates were evaluated on day 0, 1, 2, 3, 5 and 7. Immunofluorescence staining of dendritic cells(DCs), natural killer(NK) cells, macrophages, B cells, CD4+ and CD8+ T cells were measured on day 1 and 7 after renal I/RI. Survival rates dropped below 50% after day 3 in 45 minutes ischemia. Histologic analysis of ischemic kidneys revealed a significant loss of tubular architecture and infiltration of inflammatory cells. DCs, NK cells, macrophages, CD4+ and CD8+ T cells were infiltrated from a day after I/RI depending on ischemia time. Antigen presenting cells(DCs, NK cells or macrophages) and even T cells were infiltrated 24 hours post-I/RI, which is at the time of acute tubular necrosis. During the regeneration phase, not only these cells increased but B cells also appeared in more than 45 minutes ischemia. The numbers of the innate and the adaptive immune cells increased depending on ischemia as well as reperfusion time. These changes of infiltrating cells resulting from each I/RI model show that ischemic time plays a role in activating rejection related immune factors and have consequences on progression of renal disease in transplanted and native kidneys.

Role of Citrate Synthase in Acetate Utilization and Protection from Stress-Induced Apoptosis

  • Lee, Yong-Joo;Kang, Hong-Yong;Maeng, Pil Jae
    • Proceedings of the Microbiological Society of Korea Conference
    • /
    • 2008.05a
    • /
    • pp.39-41
    • /
    • 2008
  • The yeast Saccharomyces cerevisiae has been shown to contain three isoforms of citrate synthase (CS). The mitochondrial CS, Cit1, catalyzes the first reaction of the TCA cycle, i.e., condensation of acetyl-CoA and oxaloacetate to form citrate [1]. The peroxisomal CS, Cit2, participates in the glyoxylate cycle [2]. The third CS is a minor mitochondrial isofunctional enzyme, Cit3, and related to glycerol metabolism. However, the level of its intracellular activity is low and insufficient for metabolic needs of cells [3]. It has been reported that ${\Delta}cit1$ strain is not able to grow with acetate as a sole carbon source on either rich or minimal medium and that it shows a lag in attaining parental growth rates on nonfermentable carbon sources [2, 4, 5]. Cells of ${\Delta}cit2$, on the other hand, have similar growth phenotype as wild-type on various carbon sources. Thus, the biochemical basis of carbon metabolism in the yeast cells with deletion of CIT1 or CIT2 gene has not been clearly addressed yet. In the present study, we focused our efforts on understanding the function of Cit2 in utilizing $C_2$ carbon sources and then found that ${\Delta}cit1$ cells can grow on minimal medium containing $C_2$ carbon sources, such as acetate. We also analyzed that the characteristics of mutant strains defective in each of the genes encoding the enzymes involved in TCA and glyoxylate cycles and membrane carriers for metabolite transport. Our results suggest that citrate produced by peroxisomal CS can be utilized via glyoxylate cycle, and moreover that the glyoxylate cycle by itself functions as a fully competent metabolic pathway for acetate utilization in S. cerevisiae. We also studied the relationship between Cit1 and apoptosis in S. cerevisiae [6]. In multicellular organisms, apoptosis is a highly regulated process of cell death that allows a cell to self-degrade in order for the body to eliminate potentially threatening or undesired cells, and thus is a crucial event for common defense mechanisms and in development [7]. The process of cellular suicide is also present in unicellular organisms such as yeast Saccharomyces cerevisiae [8]. When unicellular organisms are exposed to harsh conditions, apoptosis may serve as a defense mechanism for the preservation of cell populations through the sacrifice of some members of a population to promote the survival of others [9]. Apoptosis in S. cerevisiae shows some typical features of mammalian apoptosis such as flipping of phosphatidylserine, membrane blebbing, chromatin condensation and margination, and DNA cleavage [10]. Yeast cells with ${\Delta}cit1$ deletion showed a temperature-sensitive growth phenotype, and displayed a rapid loss in viability associated with typical apoptotic hallmarks, i.e., ROS accumulation, nuclear fragmentation, DNA breakage, and phosphatidylserine translocation, when exposed to heat stress. Upon long-term cultivation, ${\Delta}cit1$ cells showed increased potentials for both aging-induced apoptosis and adaptive regrowth. Activation of the metacaspase Yca1 was detected during heat- or aging-induced apoptosis in ${\Delta}cit1$ cells, and accordingly, deletion of YCA1 suppressed the apoptotic phenotype caused by ${\Delta}cit1$ mutation. Cells with ${\Delta}cit1$ deletion showed higher tendency toward glutathione (GSH) depletion and subsequent ROS accumulation than the wild-type, which was rescued by exogenous GSH, glutamate, or glutathione disulfide (GSSG). Beside Cit1, other enzymes of TCA cycle and glutamate dehydrogenases (GDHs) were found to be involved in stress-induced apoptosis. Deletion of the genes encoding the TCA cycle enzymes and one of the three GDHs, Gdh3, caused increased sensitivity to heat stress. These results lead us to conclude that GSH deficiency in ${\Delta}cit1$ cells is caused by an insufficient supply of glutamate necessary for biosynthesis of GSH rather than the depletion of reducing power required for reduction of GSSG to GSH.

  • PDF