摘要 :
A field of spoken dialog systems is a rapidly growing research area because the performanceimprovement of speech technologies motivates the possibility of building systems that a humancan easily operate in order to access useful i...
展开
A field of spoken dialog systems is a rapidly growing research area because the performanceimprovement of speech technologies motivates the possibility of building systems that a humancan easily operate in order to access useful information via spoken languages. Among thecomponents in a spoken dialog system, the dialog management plays major roles such asdiscourse analysis, database access, error handling, and system action prediction. This surveycovers design issues and recent approaches to the dialog management techniques for modelingthe dialogs. We also explain the user simulation techniques for automatic evaluation of spokendialog systems.
收起
摘要 :
This paper proposes a novel user intention simulation method which is data-driven but can integrate diverse user discourse knowledge to simulate various types of user behaviors. A method of data-driven user intention modeling base...
展开
This paper proposes a novel user intention simulation method which is data-driven but can integrate diverse user discourse knowledge to simulate various types of user behaviors. A method of data-driven user intention modeling based on logistic regression is introduced in the Markov logic framework. Human dialog knowledge is designed into two layers, domain and discourse knowledge, and integrated with the data-driven model in generation time. Three types of user knowledge, i.e., cooperative, corrective and self-directing, are designed and integrated to generate behaviors of corresponding user-types. In experiments to investigate the patterns of simulated users, the approach successfully generated cooperative, corrective and self-directing user intention patterns.
收起
摘要 :
Network pruning reduces the number of parameters and computational costs of convolutional neural networks while maintaining high performance. Although existing pruning methods have achieved excellent results, they do not consider ...
展开
Network pruning reduces the number of parameters and computational costs of convolutional neural networks while maintaining high performance. Although existing pruning methods have achieved excellent results, they do not consider reconstruction after pruning in order to apply the network to actual devices. This study proposes a reconstruction process for channel-based network pruning. For lossless reconstruction, we focus on three components of the network: the residual block, skip connection, and convolution layer. Union operation and index alignment are applied to the residual block and skip connection, respectively. Furthermore, we reconstruct a compressed convolution layer by considering batch normalization. We apply our method to existing channel-based pruning methods for downstream tasks such as image classification, object detection, and semantic segmentation. Experimental results show that compressing a large model has a 1.93% higher accuracy in image classification, 2.2 higher mean Intersection over Union (mIoU) in semantic segmentation, and 0.054 higher mean Average Precision (mAP) in object detection than well-designed small models. Moreover, we demonstrate that our method can reduce the actual latency by 8.15× and 5.29× on Raspberry Pi and Jetson Nano, respectively.
收起
摘要 :
Uterine cervical cancer (CC) is a complex, multistep disease primarily linked to persistent infection with high-risk human papillomavirus (HR-HPV). However, it is widely acknowledged that HR-HPV infection alone cannot account for ...
展开
Uterine cervical cancer (CC) is a complex, multistep disease primarily linked to persistent infection with high-risk human papillomavirus (HR-HPV). However, it is widely acknowledged that HR-HPV infection alone cannot account for the formation and progression of CC. Emerging evidence suggests that the cervicovaginal microbiome (CVM) also plays a significant role in HPV-related CC. Certain bacteria, such as Fusobacterium spp., Porphyromonas, Prevotella, and Campylobacter, are currently being considered as potential microbiomarkers for HPV-positive CC. However, the composition of the CVM in CC is inconsistent; thus, further studies are needed. This review comprehensively discusses the complex interplay between HPV and the CVM in cervical carcinogenesis. It is postulated that the dynamic interaction between HPV and the CVM creates an imbalanced cervicovaginal microenvironment that triggers dysbiosis, enhances HPV persistence, and promotes cervical carcinogenesis. Moreover, this review aims to provide updated evidence on the potential role of bacteriotherapy, particularly probiotics, in the treatment of CC.
收起
摘要 :
Infrared (IR) thermography was used to measure the local heat transfer coefficients within two plate heat exchanger geometries. The chevron patterns were machined into polycarbonate and IR transparent calcium fluoride plates, both...
展开
Infrared (IR) thermography was used to measure the local heat transfer coefficients within two plate heat exchanger geometries. The chevron patterns were machined into polycarbonate and IR transparent calcium fluoride plates, both of which were electrically heated using flexible film heaters at heat fluxes up to 0.8 W cm~(-2). The test fluid was a refrigerant (HFE7100) at mass fluxes between 25 and 100 kg m~(-2) s~(-1) and qualities from 0 to 0.9. The apparatus and data reduction technique were validated by comparing the single-phase heat transfer and pressure drop data against the prediction methods from the literature. Adiabatic flow visualizations were conducted to link the flow patterns with the observed heat transfer. The frictional pressure gradient and heat transfer coefficient were compared with available correlations. It was shown that the heat transfer coefficient and the frictional pressure gradient increased with mass flux and quality. The comparison indicated the need for new prediction methods for predicting the local thermal-hydraulic performance over a wide range of operating conditions.
收起
摘要 :
The conversion of glycerol to high-value-added products via photoelectrochemical (PEC) oxidation has emerged as a promising approach for utilizing a sustainable and clean energy source with environmental and economic benefits. Mor...
展开
The conversion of glycerol to high-value-added products via photoelectrochemical (PEC) oxidation has emerged as a promising approach for utilizing a sustainable and clean energy source with environmental and economic benefits. Moreover, the energy requirement for glycerol to produce hydrogen is lower than that for pure water splitting. In this study, we propose the use of WO3 nanostructures decorated with Bi-based metal-organic frameworks (Bi-MOFs) as the photoanode for glycerol oxidation with simultaneous hydrogen production. The WO3-based electrodes selectively converted glycerol to glyceraldehyde, a high-value-added product, with remarkable selectivity. The Bi-MOF-decorated WO3 nanorods enhanced the surface charge transfer and adsorption properties, thereby improving the photocurrent density and production rate (1.53 mA/cm2 and 257 mmol/m2·h at 0.8 VRHE). The photocurrent was maintained for 10 h, ensuring stable glycerol conversion. Furthermore, at 1.2 VRHE, the average production rate of glyceraldehyde reached 420 mmol/m2·h, with a selectivity of 93.6% between beneficial oxidized products over the photoelectrode. This study provides a practical approach for the conversion of glycerol to glyceraldehyde via the selective oxidation of WO3 nanostructures and demonstrates the potential of Bi-MOFs as a promising cocatalyst for PEC biomass valorization.
收起
摘要 :
Haptic interactions play an essential role in education to enhance learning efficiency; however, haptic information for virtual educational content remains lacking. This article proposes a planar cable-driven haptic interface with...
展开
Haptic interactions play an essential role in education to enhance learning efficiency; however, haptic information for virtual educational content remains lacking. This article proposes a planar cable-driven haptic interface with movable bases that can display isotropic force feedback with maximum workspace extension on a commercial screen display. A generalized kinematic and static analysis of the cable-driven mechanism is derived by considering movable pulleys. Based on the analyses, a system including movable bases is designed and controlled to maximize the workspace subject to isotropic force exertion for the target screen area. The proposed system is evaluated experimentally as a haptic interface represented by the workspace, isotropic force-feedback range, bandwidth, Z-width, and user experiment. The results indicate that the proposed system can maximize workspace to the target rectangular area and exert isotropic force up to 94.0% of the theoretical computed one within the workspace.
收起
摘要 :
This study introduces an innovative framework that utilizes a limited number of sensors to detect chemical leaks early, enabling a timely and appropriate initial response. Consequently, it mitigates the risk of major industrial di...
展开
This study introduces an innovative framework that utilizes a limited number of sensors to detect chemical leaks early, enabling a timely and appropriate initial response. Consequently, it mitigates the risk of major industrial disasters. This framework integrates a surrogate model based on differentiable physics with a sensor placement optimization tool. Particularly noteworthy is that the surrogate model provides faster results compared to traditional CFD tools and seamlessly integrates with a CNN-based deep learning model, yielding higher-resolution outputs. This enhances the feasibility of practical application compared to conventional methods. Sensor placement optimization employs MINLP, considering various constraints to achieve more precise optimization. The proposed framework offers an innovative approach to address industrial safety concerns by enabling early chemical leak detection and facilitating timely responses, ultimately reducing risks like fires, explosions, and toxic incidents.
收起
摘要 :
Background/Aim: Cervical cancer is the fourth most common type of cancer in women worldwide and it is a major cause of cancer-related deaths in developing countries. Despite the marked reduction observed in the rates of the diseas...
展开
Background/Aim: Cervical cancer is the fourth most common type of cancer in women worldwide and it is a major cause of cancer-related deaths in developing countries. Despite the marked reduction observed in the rates of the disease as a result of screening programs, it is necessary to develop robust biomarkers that can detect the neoplastic progression early in HPV-related cervical lesions. Materials and Methods: We performed comparative mRNA sequencing from exfoliative cervical cytology samples from nine Korean women using the Illumina NovaSeq6000 platform. Each pathological tissue was matched to the corresponding cytological sample. The pathologic diagnosis was scrutinized with ancillary immunohistochemistry and was considered a confirmative (endpoint) diagnosis. The pathological diagnoses consisted of three cases of chronic cervicitis, 2 high-grade squamous intraepithelial lesions (HSILs), 2 squamous cell carcinomas in situ (CIS), and 2 invasive squamous cell carcinomas (SQCCs), respectively. Using bioinformatic analyses, differentially expressed genes (DEGs; fold change ≥1.5; p<0.05) were applied for Gene Ontology (GO), Gene Set Enrichment Analysis (GSEA), and protein-protein interaction (PPI) networks. Results: From a total of 55,882 genes, 438 DEGs were pinpointed; 282 genes were up-regulated and 156 genes down-regulated. These transcriptomic profiles were clearly divided into neoplastic (HSIL, CIS, and SQCC; ≥HSILs) and non-neoplastic lesions. The up-regulated DEGs were HIF-1a, EDN1, PIK3R3, PPP1CA and AKR1C1. GO, GSEA, and PPI network analyses showed marked associations with metabolism, proteolysis, or proteoglycan process pathways in cervical carcinogenesis. Conclusion: The transcriptomic analysis using exfoliative cervical cells was more likely representative of its corresponding histopathological diagnosis, thus emphasizing its potential utility in clinical practice. This study provides comprehensive transcriptomic network analyses for robust biomarkers that might present a high potential risk of progression to cancer in the exfoliative cervical cytology; our findings support their clinical utility for improved cervical cancer screening.
收起
摘要 :
Fiber media filtration for combined sewer overflow (CSO) treatment was evaluated in this work. Pilot-scale experiments of CSO treatment involve seven layers that form radial-type fiber media filters. The fiber media filter was cha...
展开
Fiber media filtration for combined sewer overflow (CSO) treatment was evaluated in this work. Pilot-scale experiments of CSO treatment involve seven layers that form radial-type fiber media filters. The fiber media filter was characterized using analysis techniques and exhibited extremely high hydrophobicity and negative charge. The results obtained for a sewer treatment plant at Tancheon in Seoul revealed potential for suspended solid, total organic carbon, and biochemical oxygen demand removal with efficiencies of 89.6%, 56%, and 42%, respectively. The results are consistent with the sieving effects and electrostatic repulsion between fiber media and pollutants. Head loss was induced by organic pollutants and was 10.5 cm after 20 h of operation. To mitigate the head loss induced by organic pollutants on the fiber media filter, sodium hypochlorite (NaClO) was used as a chemical backwashing agent. An NaClO concentration of 2000 mg/L with a soaking time of 24 h yielded an optimal head loss recovery of 96.2% of the initial head loss obtained for the virgin fiber media filter without media filter degradation. The result of the present study will provide practical insight and act as a technical guide for CSO treatment plant engineers.
收起