Search

Academic performance, a globally understood metric, is utilized worldwide across disparate teaching and learning environments and is regarded as a quantifiable indicator of learning gain. The ability to reliably estimate student's academic performance is important and can assist academic staff to improve the provision of support. However, it is recognized that academic performance estimation is non-trivial and affected by multiple factors, including a student's engagement with learning activities and their social, geographic, and demographic characteristics. This paper investigates the opportunity to develop reliable models for predicting student performance using Artificial Intelligence. Specifically, we propose two-step academic performance prediction using feature weighted support vector machine and artificial neural network (ANN) learning. A feature weighted SVM, where the importance of different features to the outcome is calculated using information gain ratios, is employed to perform coarse-grained binary classification (pass, P 1 , or fail, P 0 ). Subsequently, detailed score levels are divided from D to A+, and ANN learning is employed for fine-grained, multi-class training of the P 1 and P 0 classes separately. The experiments and our subsequent ablation study, which are conducted on the student datasets from two Portuguese secondary schools, have proved the effectiveness of this hybridized method. [ABSTRACT FROM AUTHOR]

This research aimed to identify quantitative trait loci (QTLs) associated with seed protein concentration in a recombinant inbred line (RIL) population of pea and aimed to validate the identified QTLs using chromosome segment-introgressed lines developed by recurrent backcrossing. PR-25, an RIL population consisting of 108 F7 bulked lines derived from a cross between CDC Amarillo (yellow cotyledon) and CDC Limerick (green cotyledon), was used in this research. The RIL population was genotyped using an Axiom 90K SNP array. A total of 10,553 polymorphic markers were used for linkage map construction, after filtering for segregation distortion and missing values. The linkage map represents 901 unique loci on 11 linkage groups which covered a map distance of 855.3 Centimorgans. Protein concentration was assessed using near-infrared (NIR) spectroscopy of seeds harvested from field trials in seven station-years in Saskatchewan, Canada, during the 2019–2021 field seasons. Three QTLs located on chromosomes 2, 3 and 5 were identified to be associated with seed protein concentration. These QTLs explained 22%, 11% and 17% of the variation for protein concentration, respectively. The identified QTLs were validated by introgression lines, developed by marker-assisted selection of backcross lines for introgression of corresponding chromosome segments (~1/4 chromosome) harboring the QTL regions. Introgression line PR-28-7, not carrying any protein-related QTLs identified in this study, was 4.7% lower in protein concentration than CDC Amarillo, the lower protein parent of PR-25 which carried one identified protein-related QTL. The SNP markers located at the peak of the three identified QTLs will be converted into breeder-friendly KASP assays, which will be used for the selection of high-protein lines from segregating populations. [ABSTRACT FROM AUTHOR]

Surface reconstruction from point clouds plays a crucial role in computer vision. The current state-of-the-art methods solve this problem by learning signed distance functions (SDFs) with ground truth distance supervisions, which are difficult to obtain. Moreover, most recent works represent each shape with a single or several latent codes, which fail to provide detailed guidance to reconstruct the local geometry. To address these issues, we propose MGSDF, a novel method for high-fidelity and fast surface reconstruction from raw point clouds. Specifically, we design a scalable representation with learnable hierarchical feature grids to capture multi-level geometric details. We introduce a self-supervised learning scheme that optimizes the SDF directly from the raw point cloud by pulling the space onto the surface. In addition, we propose a field regularization constraint on the predicted distance values and gradients on the zero-level set of SDFs for robust optimization. Our experimental results demonstrate significant improvements over the state-of-the-art in surface reconstruction from clean, noisy and varying density point clouds under widely used benchmarks. [Display omitted] • We propose MGSDF, a novel framework that employs a learnable multi-grid representation and a self-supervised schema to directly reconstruct surfaces from raw point clouds, without relying on ground truth signed distance values. • We introduce a Self-Loop loss as a field regularization constraint to provide more robust guidance to the network, resulting in improved reconstruction accuracy. • We conducted comprehensive evaluations on various benchmarks and achieved state-of-the-art results in surface reconstruction under different conditions. [ABSTRACT FROM AUTHOR]

In this paper, we propose a new semantic segmentation network called 'E-Res U-Net', to achieve better segmentation results of deep and superficial muscles in ultrasonic muscle images. This model is based on U-Net, and its structure has been modified to improve the performance of the algorithm. There are three aspects of improvement based on U-Net, including E-Res layer, dilated convolution module, and E-Res path. Additional experiments demonstrate that each designed module in our proposed network is effective, can improve the accuracy compared to the original U-Net. When compared with other algorithms which are state-of-the-art, the experimental result under the overall network structure is even more excellent. • Segmentation of deep and superficial muscles in ultrasonic muscle images. • An improved U-Net model according with three designed module is presented. • Each designed module can effectively improve the accuracy of segmentation. • Compared with other model, the segmentation effect of is significantly improved. [ABSTRACT FROM AUTHOR]

• Series of Co-doped PB particles with different Co2+ doping (Co-PB) were successfully synthesized. • The photothermal-enhanced fenton reaction of the Co 75% -PB was highlighted. • The photothermal efficiency of Co 75% -PB was calculated to be 37.6%. Series of Co-doped PB particles with different Co2+ doping (Co-PB) were synthesized and characterized. Through methylene blue degradation experiments, and the Co 75% -PB with the best catalytic performance was screened out. The photothermal efficiency of Co 75% -PB was calculated to be 37.6 %, and the degradation performance of Co 75% -PB on methylene blue can be further enhanced with photothermal effect. As a result, Co 75％ -PB nanoparticles can be potentially used for future cancer theranostics. [ABSTRACT FROM AUTHOR]

Semantic understanding is an essential research issue for many applications, such as social network analysis, collective intelligence and content computing, which tells the inner meaning of language form. Recently, Abstract Meaning Representation (AMR) is attracted by many researchers for its semantic representation ability on an entire sentence. However, due to the non-projectivity and reentrancy properties of AMR graphs, they lose some important semantic information in parsing from sentences. In this paper, we propose a general AMR parsing model which utilizes a two-stack-based transition algorithm for both Chinese and English datasets. It can incrementally parse sentences to AMR graphs in linear time. Experimental results demonstrate that it is superior in recovering reentrancy and handling arcs while is competitive with other transition-based neural network models on both English and Chinese datasets. [ABSTRACT FROM AUTHOR]

Noun phrases reflect people's understanding of the world entities and play an important role in people's language system, conceptual system and application system. With the Chinese "的(de)" structure, attributive noun phrases of the combined type can accommodate more words and syntactic structures, resulting in rich levels and complex semantic structures in Chinese sentences. Moreover, the Chinese elliptical "的(de)" structure is also of vital importance to the overall semantic understanding of the sentence. Many researches focus on rule-based models and semantic complement of "verb+的(de)" structure. To tackle these issues, we propose a general three-stage strategy utilizing neural network for the researches on all "的(de)" structure. Experimental results demonstrate that the proposed strategy is effective in boundary definition, elliptical recognition and semantic complement of "的(de)" structure. [ABSTRACT FROM AUTHOR]

This paper proposes two algorithms for signal timing optimization of single intersections, namely, microbial genetic algorithm and simulated annealing algorithm. The basis of the optimization of these two algorithms is the original timing scheme of the SCATS, and the optimized parameters are the average delay of vehicles and the capacity. Experiments verify that these two algorithms are, respectively, improved by 67.47% and 46.88%, based on the original timing scheme. [ABSTRACT FROM AUTHOR]

Short text categorization is a crucial issue to many applications, e.g., Information Retrieval, Question-Answering System, MRI Database Construction and so forth. Many researches focus on data sparsity and ambiguity issues in short text categorization. To tackle these issues, we propose a novel short text categorization strategy based on abundant representation, which utilizes Bi-directional Recurrent Neural Network(Bi-RNN) with Long Short-Term Memory(LSTM) and topic model to catch more contextual and semantic information. Bi-RNN enriches contextual information, and topic model discovers more latent semantic information for abundant text representation of short text. Experimental results demonstrate that the proposed model is comparable to state-of-the-art neural network models and method proposed is effective. [ABSTRACT FROM AUTHOR]