41 research outputs found
Causal-Paced Deep Reinforcement Learning
No full textDesigning effective task sequences is crucial for curriculum reinforcement learning (CRL), where agents must gradually acquire skills by training on intermediate tasks. A key challenge in CRL is to identify tasks that promote exploration, yet are similar enough to support effective transfer. While recent approach suggests comparing tasks via their Structural Causal Models (SCMs), the method requires access to ground-truth causal structures, an unrealistic assumption in most RL settings. In this work, we propose Causal-Paced Deep Reinforcement Learning (CP-DRL), a curriculum learning framework aware of SCM differences between tasks based on interaction data approximation. This signal captures task novelty, which we combine with the agent’s learnability, measured by reward gain, to form a unified objective. Empirically, CP-DRL outperforms existing curriculum methods on the Point Mass benchmark, achieving faster convergence and higher returns. CP-DRL demonstrates reduced variance with comparable final returns in the Bipedal Walker-Trivial setting, and achieves the highest average performance in the Infeasible variant. These results indicate that leveraging causal relationships between tasks can improve the structure-awareness and sample efficiency of curriculum reinforcement learning. We provide the full implementation of CP-DRL to facilitate the reproduction of our main results at \url{https://github.com/Cho-Geonwoo/CP-DRL}
Temperature-Dependent Plasmonic Response of Graphene Nanoresonators
No full textGraphene plasmons have attracted enormous research interest due to their dynamic tunability and the extreme field confinement they provide. However, despite their popularity, most studies revolving around graphene plasmons have been restricted to room temperature, leaving unconsidered important tunability knob. In this work, we experimentally investigate the temperature-dependent plasmonic properties of graphene nanoresonators with varying widths on SiO2 substrate by infrared transmission spectroscopy. As temperature drops from 300 to 100 K, the intensity of the graphene plasmon resonance peak increases up to 76%, and the amount of enhancement decreases with increasing carrier concentration and decreasing resonator width. We attribute the enhancement of graphene plasmon resonance to an additional hole doping of Delta p = 1.37 X 10(1)(2) cm(-2) associated with cooling and reduced plasmon damping due to the suppression of phonon-mediated scattering channels. Our results uncover the significance of temperature effects that can be exploited in graphene-based tunable plasmonic devices operating at low temperatures.
Use of Measurement Tools to Validate the Health Effects of Forest Healing Programs: A Qualitative Analysis
No full textResearch is increasingly focusing on the health-promoting effects of forest healing programs. A number of different health indicators are in use, necessitating the establishment of standardized health measurement tools and protocols for forest healing. Also, it is necessary to improve the indicators and protocols by incorporating the opinions of program participants and instructors, so we conducted a qualitative analysis based on focus group interviews (FGIs) and in-depth interviews (IDIs). We collected opinions through interviews conducted for about 1 h with 32 participants in the pilot study and three instructors of the forest healing program. We utilized the MAXQDA program, commonly employed for qualitative research, such as coding and analyzing interview transcripts and literature reviews, as part of the qualitative research process. Three researchers coded and categorized the data, and the first author and corresponding author performed the final coding and categorization. Opinions on the five mental health questionnaires, three physical health measures, and exercise behavior measures used in the forest healing program were solicited. Opinions on the measurement protocol were also solicited. Participants faced challenges in completing the mental health questionnaire due to inappropriate terminology, difficulty in providing truthful responses due to repetitive questions, and the complexity of answering exercise-type questions due to the length of the survey and the absence of clear examples. It was identified that improvements are needed in the future. Some participants commented on the need to measure blood circulation and short-term health changes, and others noted that performing measurements in large groups was difficult, such that there was a need to introduce a measurement protocol for groups. This study is the first to qualitatively evaluate the validity of health measurement tools associated with forest therapy programs. It can contribute to the establishment of standardized health indicators and protocols, as well as serve as a valuable reference for selecting measurement tools to evaluate the effectiveness of forest healing interventions
Biological Effects of Double-Layered Hydroxyapatite and Zirconium Oxide Depositions on Titanium Surfaces
No full textMin-Kyung Ji,1,* Yaerim Chun,2,* Geonwoo Jeong,3 Hyun-Seung Kim,4 Won-Jae Kim,5 Je-Hwang Ryu,6 Hoonsung Cho,3 Hyun-Pil Lim1,2 1Dental 4D Research Center, Chonnam National University, Gwangju, Republic of Korea; 2Department of Prosthodontics, School of Dentistry, Chonnam National University, Gwangju, Republic of Korea; 3Department of Materials Science and Engineering, Chonnam National University, Gwangju, Republic of Korea; 4KJ Meditech Co., Ltd, Gwangju, Republic of Korea; 5Department of Oral Physiology, School of Dentistry, Stem Cell Secretome Research Center, Chonnam National University, Gwangju, Republic of Korea; 6Department of Pharmacology and Dental Therapeutics, School of Dentistry, Chonnam National University, Gwangju, Republic of Korea*These authors contributed equally to this workCorrespondence: Hyun-Pil Lim, Department of Prosthodontics, School of Dentistry, Chonnam National University, Gwangju, 61186, Republic of Korea, Tel +82-10-2645-7528, Fax +82-62-530-5577, Email [email protected] Hoonsung Cho, Department of Materials Science and Engineering, Chonnam National University, Gwangju, 61186, Republic of Korea, Tel/Fax +82-62-530-1717, Email [email protected]: This study aimed to confirm the synergy effect of these two materials by evaluating osteoblast and antibacterial activity by applying a double-layered hydroxyapatite(HA) zirconium oxide(ZrO2) coating to titanium.Methods: The specimens used in this study were divided into four groups: a control group (polished titanium; group T) and three experimental groups: Group TH (RF magnetron sputtered HA deposited titanium), Group Z (ZrO2 ALD deposited titanium), and Group ZH (RF magnetron sputtered HA and ZrO2 ALD deposited titanium). The adhesion of Streptococcus mutans (S.mutans) to the surface was assessed using a crystal violet assay. The adhesion, proliferation, and differentiation of MC3T3-E1 cells, a mouse osteoblastic cell line, were assessed through a WST-8 assay and ALP assay.Results: Group Z showed a decrease in the adhesion of S. mutans (p < 0.05) and an improvement in osteoblastic viability (p < 0.0083). Group TH and ZH showed a decrease in adhesion of S. mutans (p < 0.05) and an increase in osteoblastic cell proliferation and cell differentiation (p < 0.0083). Group ZH exhibited the highest antibacterial and osteoblastic differentiation.Conclusion: In conclusion double-layered HA and ZrO2 deposited on titanium were shown to be more effective in inhibiting the adhesion of S. mutans, which induced biofilm formation, and increasing osteoblastic differentiation involved in osseointegration by the synergistic effect of the two materials.Keywords: zirconium oxide, hydroxyapatite, atomic layer deposition, radio frequency magnetron sputter, osteoblast activity, antibacterial effec
Snapshot-Based Visible-Near Infrared Multispectral Imaging for Early Screening of Heat Injury during Growth of Chinese Cabbage
No full textHeat stress in particular can damage physiological processes, adaptation, cellular homeostasis, and yield of higher plants. Early detection of heat stress in leafy crops is critical for preventing extensive loss of crop productivity for global food security. Thus, this study aimed to evaluate the potential of a snapshot-based visible-near infrared multispectral imaging system for detecting the early stage of heat injury during the growth of Chinese cabbage. Two classification models based on partial least squares-discriminant analysis (PLS-DA) and least-squares support vector machine (LS-SVM) were developed to identify heat stress. Various vegetation indices (VIs), including the normalized difference vegetation index (NDVI), red-edge ratio (RE/R), and photochemical reflectance index (PRI), which are closely related to plant heat stress, were acquired from sample images, and their values were compared with the developed models for the evaluation of their discriminant performance of developed models. The highest classification accuracies for LS-SVM, PLS-DA, NDVI, RE/R, and PRI were 93.6%, 92.4%, 72.5%, 69.6%, and 58.1%, respectively, without false-positive errors. Among these methods for identifying plant heat stress, the developed LS-SVM and PLS-DA models showed more reliable discriminant performance than the traditional VIs. This clearly demonstrates that the developed models are much more effective and efficient predictive tools for detecting heat stress in Chinese cabbage in the early stages compared to conventional methods. The developed technique shows promise as an accurate and cost-effective screening tool for rapid identification of heat stress in Chinese cabbage
itKD: Interchange Transfer-based Knowledge Distillation for 3D Object Detection
No full textPoint-cloud based 3D object detectors recently have achieved remarkable
progress. However, most studies are limited to the development of network
architectures for improving only their accuracy without consideration of the
computational efficiency. In this paper, we first propose an autoencoder-style
framework comprising channel-wise compression and decompression via interchange
transfer-based knowledge distillation. To learn the map-view feature of a
teacher network, the features from teacher and student networks are
independently passed through the shared autoencoder; here, we use a compressed
representation loss that binds the channel-wised compression knowledge from
both student and teacher networks as a kind of regularization. The decompressed
features are transferred in opposite directions to reduce the gap in the
interchange reconstructions. Lastly, we present an head attention loss to match
the 3D object detection information drawn by the multi-head self-attention
mechanism. Through extensive experiments, we verify that our method can train
the lightweight model that is well-aligned with the 3D point cloud detection
task and we demonstrate its superiority using the well-known public datasets;
e.g., Waymo and nuScenes.Comment: Accepted at CVPR 202
Centrifugal Lithography: Self-Shaping of Polymer Microstructures Encapsulating Biopharmaceutics by Centrifuging Polymer Drops
No full textPolymeric microstructures encapsulating biopharmaceutics must be fabricated in a controlled environment to preserve the biological activity. There is increasing demand for simple methods designed to preserve the biological activity by utilizing the natural properties of polymers. Here, the paper shows that centrifugal lithography (CL) can be used for the fabrication of such microstructures in a single centrifugation, by engineering the self-shaping properties of hyaluronic acid (HA). In this method, HA drops are self-shaped into hourglass-microstructures to produce two dissolving microneedles (DMN), which facilitate transdermal delivery via implantation on the skin. In addition, tuberculin purified protein derivatives are encapsulated into HA DMNs under refrigerated conditions (4 °C) during CL. Therefore, the tuberculin skin test (TST) with the DMNs indicates minimal damage, as opposed to the case of TST with traditional hypodermic needles. These findings on the fabrication of polymeric microstructures with biopharmaceutics may trigger the development of various biomedical devices and therapies.restrictio
Effects of a Forest Therapy Program on Physical Health, Mental Health, and Health Behaviors
No full text(1) Background: Although interest in the health-promoting effects of forest therapy is increasing, few researchers have investigated the mid-long-term impact of such therapy on health indicators or exercise behaviors. We explored changes in physical health, mental health, and exercise behaviors 1, 2, and 4 weeks after a forest therapy program concluded. We sought to establish a solid foundation for such programs and a standardized evaluation system. (2) Method: We measured the blood pressure and heart rate variability of 99 adults before and after participation in a forest therapy program. We used the State-Trait Anxiety Inventory to assess anxiety, the Beck Depression Inventory to evaluate both anxiety and depression, the Profile of Mood States to explore mood, the Euro-Quality of Life-5 Dimension scale to assess the overall quality of life, and the Positive and Negative Effect Schedule to measure positive and negative mood. We employed the Global Physical Activity Questionnaire to determine exercise time, intensity, and changes in exercise type before the program and 1–4 weeks after program completion. (3) Results: Anxiety, depression, mood, quality of life, heart rate, and blood pressure control improved significantly after the program. The reduced depression and increased medium-intensity exercise time persisted for 1, 2, and 4 weeks after the end of the program. (4) Conclusions: We tracked various health indicators and clearly distinguished those that were useful in the short term from those more appropriate for evaluation in the long term. This is the first report to show that a forest therapy program affects exercise behavior; this suggests that health behaviors should be continuously tracked
Bandwidth-Effective DRAM Cache for GPUs with Storage-Class Memory
No full textWe propose overcoming the memory capacity limitation of GPUs with
high-capacity Storage-Class Memory (SCM) and DRAM cache. By significantly
increasing the memory capacity with SCM, the GPU can capture a larger fraction
of the memory footprint than HBM for workloads that oversubscribe memory,
achieving high speedups. However, the DRAM cache needs to be carefully designed
to address the latency and BW limitations of the SCM while minimizing cost
overhead and considering GPU's characteristics. Because the massive number of
GPU threads can thrash the DRAM cache, we first propose an SCM-aware DRAM cache
bypass policy for GPUs that considers the multi-dimensional characteristics of
memory accesses by GPUs with SCM to bypass DRAM for data with low performance
utility. In addition, to reduce DRAM cache probes and increase effective DRAM
BW with minimal cost, we propose a Configurable Tag Cache (CTC) that repurposes
part of the L2 cache to cache DRAM cacheline tags. The L2 capacity used for the
CTC can be adjusted by users for adaptability. Furthermore, to minimize DRAM
cache probe traffic from CTC misses, our Aggregated Metadata-In-Last-column
(AMIL) DRAM cache organization co-locates all DRAM cacheline tags in a single
column within a row. The AMIL also retains the full ECC protection, unlike
prior DRAM cache's Tag-And-Data (TAD) organization. Additionally, we propose
SCM throttling to curtail power and exploiting SCM's SLC/MLC modes to adapt to
workload's memory footprint. While our techniques can be used for different
DRAM and SCM devices, we focus on a Heterogeneous Memory Stack (HMS)
organization that stacks SCM dies on top of DRAM dies for high performance.
Compared to HBM, HMS improves performance by up to 12.5x (2.9x overall) and
reduces energy by up to 89.3% (48.1% overall). Compared to prior works, we
reduce DRAM cache probe and SCM write traffic by 91-93% and 57-75%,
respectively.Comment: Published in 2024 IEEE International Symposium on High-Performance
Computer Architecture (HPCA'24
Utilizing SIFT-MS and GC-MS for Phytoncide Assessment in Phytotron: Implications for Indoor Forest Healing Programs
No full textThis study addresses the growing need for phytoncide studies, driven by the demand to design indoor forest healing programs, including virtual reality experiences, for patients unable to visit actual forests. Previous studies have struggled to establish consistent phytoncide emission patterns in outdoor forest environments owing to varying microclimates and abiotic factors. In addition, the traditional gas chromatography–mass spectroscopy (GC-MS) method presents field measurement challenges, whereas the selected ion flow tube (SIFT)-MS method offers improved efficiency. This study concentrated on a controlled phytotron environment and compared the GC-MS and SIFT-MS findings, revealing similar emission trends with slightly higher SIFT-MS concentrations. Daily phytoncide emissions fluctuated with light intensity and abiotic stressors. Both methods consistently detected pinenes, primarily emitted by Pinus strobus L. seedlings, in the phytotron. Statistical analysis confirmed the compatibility between GC-MS and SIFT-MS results, supporting the use of SIFT-MS for forest phytoncide assessment. In the second phase, the phytoncide emissions were assessed indoors, outdoors, and in the phytotron, highlighting the superiority of the phytotron under controlled conditions. Despite certain limitations, this study underscores the value of phytotron-based measurements for indoor forest healing programs and the potential adoption of SIFT-MS in future field-based phytoncide research
