Angular Shrinkage Research Articles

An angular shrinkage BERT model for few-shot relation extraction with none-of-the-above detection

Few-shot relation extraction aims to solve the problem of insufficient annotated data in relation extraction tasks. Through the comparison between samples, few-shot relation extraction achieves lower-cost relation classification. However, most existing methods only do classification within the scope of enumerated relations. For one of the main challenges faced by the application of few-shot relation extraction-the recognition of the none-of-the-above instances, there has been few works on it. In this paper, we propose an angular shrinkage BERT model for the few-shot relation extraction task with none-of-the-above detection, which uses an additive angular loss to enlarge the margins of different classes in the feature space, and obtain highly discriminative features to improve the recognition ability for none-of-the-above instances. Meanwhile, we present a two-stage training strategy to enhance the stability of the performance. We evaluate our model on the most used few-shot relation extraction dataset FewRel. Experimental results show that our approach outperforms previous sentence-pair methods in scenarios containing none-of-the-above instances, and also achieves improvements on the traditional few-shot relation extraction task compared with our baseline model.

Shrinkage pattern assessment for black pine juvenile wood delimitation

Juvenile wood is associated with shrinkage and a high probability of warping and checks in end products, reducing their value and even causing them to be rejected for structural purposes. This work studies the variations in yearly shrinkage caused when wood passes from green to dry state. The methodology proposed is based on the juvenile wood (JW) location using annual shrinkage variations. Pinus nigra Arn. grown using different silviculture techniques was chosen for study because its wood is characterized by severe shrinkage problems. This characteristic is analyzed in the radial and angular directions, of which the latter is a different methodology for evaluating warping problem rather than the commonly used ratio of anisotropy. It takes into account the annual measurements obtained from a disc cut from the tree trunk at various heights. The results were compared with the segregated JW and mature wood (MW) area obtained by radial densitometry pattern variation methodology. The results show that radial shrinkage is higher near the pith but stabilizes earlier (3–7 rings) than angular shrinkage (9–11 rings). Nevertheless, the opposite initial radial pattern occurs in angular shrinkage, which is low near the pith but increases annually towards the bark until it stabilizes. No clear effect of height in the tree when comparing the radial shrinkage patterns obtained along the bole can be defined. Silviculture practices have no significant influence on shrinkage percentage. Shrinkage and densitometry methodologies show differences in the quantification of JW volume percentage between JW and MW estimation. Results show that radial shrinkage pattern methodology, which is associated with warping problems, classifies a higher tree volume as good quality for end uses in comparison to the use of radial densitometry pattern methodology. On the other hand, if classified by radial densitometry pattern variation, warping problems associated with JW should not be a problem when sawn.

A Novel Method for Welding Residual Deformations Prediction

Prediction of welding deformations for large structure is limited by computational reasons. Nowadays, experience plays an important role in minimizing distortions and the cost of straightening is considered as intrinsic in the process. A fast but reliable simplified method could increase productivity. On one of the factors significantly influence the welding residual distortions is the welding sequence at the global assembly scale and the presence of clamps.A non linear extension of the Virtual Weld Bead Method which is based on the concepts of the Inherent Strain Method and Strain and Direct Boundary has been developed. Auxiliary elements have been introduced to apply an equivalent load on non linear elements introducing transversal shrinkage as well as angular shrinkage in this first version. A dedicated inverse analysis procedure has been developed to calibrate the equivalent load to be applied and an experimental validation has been performed. The first part of this validation is present focusing on a single welded joint with different clamping conditions, the effect of clamps has been correctly predicted.

Analysis of angular shrinkage of fused filament fabricated poly-lactic-acid prints and its relationship with other process parameters

Undoubtedly, fused filament fabrication (FFF), a class of additive manufacturing has established as one of the most efficient process for the development of a wide range of functional/non-functional products. However, the process suffers from poor geometrical and mechanical characteristics that are limiting its potential for the fabrication of sophisticated parts. In the present work, an effort has been made to investigate the effect of build orientation angle, head scan speed, and layer thickness on the shrinkage of resulting poly-lactic-acid (PLA) prints by using design of experimentation and machine learning based statistical methodologies. Firstly, a total of 20 test samples, with repletion of 5, were prepared by using an open source based commercial fused filament fabrication (FFF) setup and then investigated for their dimensional accuracies by using Coordinate Measuring Machine (CMM). Finally, the obtained data was investigated, statistically, in order to find out the significance of selected process variables on resulted dimensional features. The regression models developed for the prediction of shrinkage by using Statistical and artificial neural network were possessed R2 ∼ 97.66% and ∼ 97.78%, respectively. Further, scanning electron micrographic analysis has been carried out to observe the changes occurred in the geometrical features of the produced prints. Overall, the study concluded that rate of angular shrinkage increased with an increase in the build orientation angle and decrease in the layer thickness & head scan speed.

Vaginal cytology, vaginoscopy and progesterone profile: breeding tools in bitches

The exfoliative vaginal cytology, vaginoscopic examination of vaginal mucosa and progesterone profiles were recorded in an attempt to identify the ideal time of breeding in bitches. A total of 18 anestrus bitches were selected and divided into 03 groups (Control, CABG and eCG groups). The bitches in control group did not receive any treatment and exhibited estrus. The estrus was induced with Cabergoline (CABG) and equine Chorionic Gonadotropin (eCG) in the other two groups of bitches. In control group, higher percentage of superficial cells (89.94 ± 0.64) and lower percentage of intermediate (7.30 ± 0.77) and parabasal cells (2.76 ± 0.30) were characteristic vaginal cytological changes during estrus. Vaginoscopic examination of CABG group of bitches revealed that the vaginal mucus was creamy and paper white with angular shrinkage during estrus. In eCG group of bitches, the plasma progesterone concentration was 1.55 ± 0.19 ng/ml on day 8.00 ± 0.71 of proestrus. The conception rates were 66.66, 83.33 and 83.33 per cent in Control, Cabergoline and eCG groups, respectively. The litter size varied from 3.50 + 1.12 to 4.83 + 0.83 in the three groups.

Welding Shrinkage and Deflections – A Thermo-Mechanical Computational Model for Spatial Thin-Walled Structures

A stationary thermo-mechanical finite-element model estimates the deflections of spatial thin-walled structures in welding. The model addresses longitudinal, lateral and angular shrinkage due to fillet joints. A strutctural shell model represents the thin-walled structure. Tests at simplified welding joints serve to calibrate parameters in the computational model. The approach uses an empirical data-base of material properties of shipbuilding steels for temperatures up to 700°. The assembly simulation from stiffened panel to spatial ship structure considers previous thermal deflections of structural parts as initial state for the next assembly step. The computational model is validated using experiments of the joining process for test specimens similar to shipbuilding structures.

Welding Distortion Minimization for an Aluminum Alloy Extruded Beam Structure Using a 2D Model

Optimization of the weld sequence of a sub-assembly composed of thin walled aluminum alloy extruded beams is investigated and presented. The main factor considered is the quality of the assembly after welding, which is measured by the deformation behavior at pre-defined critical locations. The aluminum alloy extruded beam structure is simplified by a 2-D beam element model. Our methodology consists of applying pre-estimated angular shrinkages for each welding step thus eliminating use of a complex nonlinear transient analysis which would require consideration of thermo-mechanical interactions and plasticity. Two distortion modes (angular shrinkage and tilting shrinkage) are investigated and applied to model welding distortion. Different criteria for minimization of welding distortion have been investigated, such as overall deformation or weighted deformation with emphasis on some critical areas. A composite quality index is formulated, which is a weighted measure of critical deformation and the overall deformation. For simulating welding deformation and the role of sequences, some weld sequences were selected heuristically and they were simulated in ANSYS 5.2. Based on the quality index of these sequences, some were reproduced while mutating other portions of these sequences to find better sequences. Finally, deformation data for each node and weighted measures of deformation for considered sequences are presented, and final deformed shapes for different sequences, maximum principal stress, and bending moment across the beams are shown graphically.