503 research outputs found
Inae Sharon Lee, Violin; Junsoo Park, Violin
Get PDFSonata for Violin and Piano in D minor, Op. 75 / Camille Saint Saens; Por Una Cabeza / Carlos Garde
Playing with two fires at the border: Geopolitics of wildfires in the inter-Korean border
No full textAnimeCeleb: Large-Scale Animation CelebHeads Dataset for Head Reenactment
No full textWe present a novel Animation CelebHeads dataset (AnimeCeleb) to address an animation head reenactment. Different from previous animation head datasets, we utilize a 3D animation models as the controllable image samplers, which can provide a large amount of head images with their corresponding detailed pose annotations. To facilitate a data creation process, we build a semi-automatic pipeline leveraging an open 3D computer graphics software with a developed annotation system. After training with the AnimeCeleb, recent head reenactment models produce high-quality animation head reenactment results, which are not achievable with existing datasets. Furthermore, motivated by metaverse application, we propose a novel pose mapping method and architecture to tackle a cross-domain head reenactment task. During inference, a user can easily transfer one's motion to an arbitrary animation head. Experiments demonstrate an usefulness of the AnimeCeleb to train animation head reenactment models, and the superiority of our crossdomain head reenactment model compared to state-of-the-art methods. Our dataset and code are available at this url
Empirical investigation of occupant-centric thermal comfort in hotel guestrooms
No full textGiven the importance of occupant satisfaction with thermal comfort in hotel accommodations, previous studies have assessed thermal comfort using either local or global standards. In a hotel catering to diverse guests, a single standard may constrain the formulation of an optimal strategy for improving occupant satisfaction. To address this challenge, this study empirically investigates occupant-centric thermal comfort in hotel guestrooms located in a hot and humid climatic region. To this end, a living-lab-based case study was conducted on a hotel guestroom in Hong Kong, wherein two types of thermal comfort improvement strategies were devised from both local and global perspectives. First, to improve thermal comfort in line with the local standard, it was determined that an ancillary fan coil unit (FCU) energy consumption of 6.0 kW h (approximately 34 % of the existing energy consumption) was required. Second, to improve thermal comfort in line with the global standard, an ancillary FCU energy consumption of 9.4 kW h (approximately 53 % of the existing energy consumption) was necessary. Consequently, in a hot and humid climatic region, the distribution of thermal comfort may vary depending on the type of guest (e.g., domestic or foreign), and ancillary cooling energy expenditure may be incurred to enhance a guest's thermal comfort. The proposed approach is expected to contribute to the creation of occupant-tailored thermal comfort by considering an individual's thermal acclimation to a specific climatic region from both local and global perspectives.
Systems biology for reverse aging
No full textCellular senescence is an irreversible and permanent cell cycle arrest in response to internal and external stresses. Its unresponsiveness to growth factor signals distinguishes it from a potentially reversible state, quiescence. Cellular senescence can inhibit tumor development by blocking proliferation of damaged cells, but as senescent cells become accumulated in a tissue, they can contribute to the promotion of agerelated diseases such as cancer by secreting inflammatory cytokines [1]. © 2021 Cho et al. This is an open access article distributed under the terms of the Creative Commons Attribution License (CC BY 3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited
DFX: A Low-latency Multi-FPGA Appliance for Accelerating Transformer-based Text Generation
No full textTransformer is a deep learning language model widely used for natural language processing (NLP) services in datacenters. Among transformer models, Generative Pretrained Transformer (GPT) has achieved remarkable performance in text generation, or natural language generation (NLG), which needs the processing of a large input context in the summarization stage, followed by the generation stage that produces a single word at a time. The conventional platforms such as GPU are specialized for the parallel processing of large inputs in the summarization stage, but their performance significantly degrades in the generation stage due to its sequential characteristic. Therefore, an efficient hardware platform is required to address the high latency caused by the sequential characteristic of text generation. In this paper, we present DFX, a multi-FPGA acceleration appliance that executes GPT-2 model inference end-to-end with low latency and high throughput in both summarization and generation stages. DFX uses model parallelism and optimized dataflow that is model-and-hardware-aware for fast simultaneous workload execution among devices. Its compute cores operate on custom instructions and provide GPT-2 operations end-to-end. We implement the proposed hardware architecture on four Xilinx Alveo U280 FPGAs and utilize all of the channels of the high bandwidth memory (HBM) and the maximum number of compute resources for high hardware efficiency. DFX achieves 5.58 × speedup and 3.99 × energy efficiency over four NVIDIA V100 GPUs on the modern GPT-2 model. DFX is also 8.21 × more cost-effective than the GPU appliance, suggesting that it is a promising solution for text generation workloads in cloud datacenters
Coloring With Limited Data: Few-Shot Colorization via Memory Augmented Networks
No full textDespite recent advancements in deep learning-based automatic colorization, they are still limited when it comes to few-shot learning. Existing models require a significant amount of training data. To tackle this issue, we present a novel memory-augmented colorization model MemoPainter that can produce high-quality colorization with limited data. In particular, our model is able to capture rare instances and successfully colorize them. Also, we propose a novel threshold triplet loss that enables unsupervised training of memory networks without the need for class labels. Experiments show that our model has superior quality in both few-shot and one-shot colorization tasks
Swd2/Cps35 determines H3K4 tri-methylation via interactions with Set1 and Rad6
No full textBackground Histone H3K4 tri-methylation (H3K4me3) catalyzed by Set1/COMPASS, is a prominent epigenetic mark found in promoter-proximal regions of actively transcribed genes. H3K4me3 relies on prior monoubiquitination at the histone H2B (H2Bub) by Rad6 and Bre1. Swd2/Cps35, a Set1/COMPASS component, has been proposed as a key player in facilitating H2Bub-dependent H3K4me3. However, a more comprehensive investigation regarding the relationship among Rad6, Swd2, and Set1 is required to further understand the mechanisms and functions of the H3K4 methylation.,Results We investigated the genome-wide occupancy patterns of Rad6, Swd2, and Set1 under various genetic conditions, aiming to clarify the roles of Set1 and Rad6 for occupancy of Swd2. Swd2 peaks appear on both the 5' region and 3' region of genes, which are overlapped with its tightly bound two complexes, Set1 and cleavage and polyadenylation factor (CPF), respectively. In the absence of Rad6/H2Bub, Set1 predominantly localized to the 5' region of genes, while Swd2 lost all the chromatin binding. However, in the absence of Set1, Swd2 occupancy near the 5' region was impaired and rather increased in the 3' region.,Conclusions This study highlights that the catalytic activity of Rad6 is essential for all the ways of Swd2's binding to the transcribed genes and Set1 redistributes the Swd2 to the 5' region for accomplishments of H3K4me3 in the genome-wide level.,
Development of inferential sensor and real-time optimizer for a vacuum distillation unit by recurrent neural network modeling of time series data
No full textIn lube base oil production, a vacuum distillation unit (VDU) is a crucial unit operation which must be controlled tightly to meet final product specifications. To address this requirement, model-based optimal control can be employed, which requires an accurate model relating to the operation and quality variables involved. However, the mathematical modeling of VDU is hampered by the lack of process knowledge and the limited online data acquisition capability for key quality variables. As a recourse, data-based modeling can be attempted to develop inferential sensors and controllers but one is faced with the problem of temporal credit assignment due to the complex characteristics of the VDU process dynamics, e.g., varying time-lags among the variables. To overcome this difficulty, the use of a stacked RNN structure with sequence length is proposed for the prediction of kinematic viscosity. In addition, regression models for the product's distillation temperatures are constructed by testing a large number of regression model structures. Then, these quality predictors are applied in developing a real-time optimizer which adjusts key operating variables in order to satisfy the product specifications despite various disturbances. Accuracies of the soft sensor and the optimization model are evaluated, first offline and then, online through a commercial plant test performed at SK Ulsan Refinery Complex in Korea. The overall approach and framework are general and can be used for the development of inferential sensors and controllers in other similar situations.
- …
