Accumulation of Corresponding Colours Under Extreme Chromatic Illuminations and modification of CAM16

The goal of this study was to investigate the chromatic adaptation under extreme chromatic lighting conditions using the magnitude estimation method. The locations of the lightings on CIE1976 u′v′ plane were close to the spectrum locus, so the colour purity was far beyond the previous studies, and the data could test the limitations of the existing models. Two psychophysical experiments were carried out, and 1,470 estimations of corresponding colours were accumulated. The results showed that CAT16 gave a good prediction performance for all the chromatic lightings except for blue lighting, and the degree of adaptation was relatively high, that is, D was close to 1. The prediction for blue lightings was modified, the results showed the performance of CAM16 could be improved by correcting the matrix instead of the D values.

HELP-DKT: An Interpretable Cognitive Model of How Students Learn Programming Based on Deep Knowledge Tracing

Student cognitive models are playing an essential role in intelligent online tutoring for programming courses. These models capture students' learning interactions and store them in the form of a set of binary responses, thereby failing to utilize rich educational information in the learning process. Moreover, the recent development of these models has been focused on improving the prediction performance and tended to adopt deep neural networks in building the end-to-end prediction frameworks. Although this approach can provide an improved prediction performance, it may also cause difficulties in interpreting the student's learning status, which is crucial for providing personalized educational feedback. To address this problem, this paper provides an interpretable cognitive model named HELP-DKT, which can infer how students learn programming based on deep knowledge tracing. HELP-DKT has two major advantages. First, it implements a feature-rich input layer, where the raw codes of students are encoded to vector representations, and the error classifications as concept indicators are incorporated. Second, it can infer meaningful estimation of student abilities while reliably predicting future performance. The experiments confirm that HELP-DKT can achieve good prediction performance and present reasonable interpretability of student skills improvement. In practice, HELP-DKT can personalize the learning experience of novice learners.

Vis-NIR Hyperspectral Imaging for Online Quality Evaluation during Food Processing: A Case Study of Hot Air Drying of Purple-Speckled Cocoyam (Colocasia esculenta (L.) Schott)

In this study, hyperspectral imaging (HSI) and chemometrics were implemented to develop prediction models for moisture, colour, chemical and structural attributes of purple-speckled cocoyam slices subjected to hot-air drying. Since HSI systems are costly and computationally demanding, the selection of a narrow band of wavelengths can enable the utilisation of simpler multispectral systems. In this study, 19 optimal wavelengths in the spectral range 400–1700 nm were selected using PLS-BETA and PLS-VIP feature selection methods. Prediction models for the studied quality attributes were developed from the 19 wavelengths. Excellent prediction performance (RMSEP < 2.0, r2P > 0.90, RPDP > 3.5) was obtained for MC, RR, VS and aw. Good prediction performance (RMSEP < 8.0, r2P = 0.70–0.90, RPDP > 2.0) was obtained for PC, BI, CIELAB b*, chroma, TFC, TAA and hue angle. Additionally, PPA and WI were also predicted successfully. An assessment of the agreement between predictions from the non-invasive hyperspectral imaging technique and experimental results from the routine laboratory methods established the potential of the HSI technique to replace or be used interchangeably with laboratory measurements. Additionally, a comparison of full-spectrum model results and the reduced models demonstrated the potential replacement of HSI with simpler imaging systems.

An Experimental Evaluation of Similarity-Based and Embedding-Based Link Prediction Methods on Graphs

The task of inferring missing links or predicting future ones in a graph based on its current structure is referred to as link prediction. Link prediction methods that are based on pairwise node similarity are well-established approaches in the literature and show good prediction performance in many realworld graphs though they are heuristic. On the other hand, graph embedding approaches learn lowdimensional representation of nodes in graph and are capable of capturing inherent graph features, and thus support the subsequent link prediction task in graph. This paper studies a selection of methods from both categories on several benchmark (homogeneous) graphs with different properties from various domains. Beyond the intra and inter category comparison of the performances of the methods, our aim is also to uncover interesting connections between Graph Neural Network(GNN)- based methods and heuristic ones as a means to alleviate the black-box well-known limitation.

An Experimental Evaluation of Similarity-Based and Embedding-Based Link Prediction Methods on Graphs

The task of inferring missing links or predicting future ones in a graph based on its current structure is referred to as link prediction. Link prediction methods that are based on pairwise node similarity are well-established approaches in the literature and show good prediction performance in many realworld graphs though they are heuristic. On the other hand, graph embedding approaches learn lowdimensional representation of nodes in graph and are capable of capturing inherent graph features, and thus support the subsequent link prediction task in graph. This paper studies a selection of methods from both categories on several benchmark (homogeneous) graphs with different properties from various domains. Beyond the intra and inter category comparison of the performances of the methods, our aim is also to uncover interesting connections between Graph Neural Network(GNN)- based methods and heuristic ones as a means to alleviate the black-box well-known limitation.

A Prospective Study of Liver Regeneration After Radiotherapy Based on a New (Su’S) Target Area Delineation

BackgroundIn this study, we designed a new (Su’S) target area delineation to protect the normal liver during liver regeneration and prospectively evaluate liver regeneration after radiotherapy, as well as to explore the clinical factors of liver regeneration and established a model and nomogram.MethodsThirty patients treated with preoperative downstaging radiotherapy were prospectively included in the training cohort, and 21 patients treated with postoperative adjuvant radiotherapy were included in the validation cohort. The cut-off points of each optimal predictor were obtained using receiver-operating characteristic analysis. A model and nomogram for liver regeneration after radiotherapy were developed and validated.ResultsAfter radiotherapy, 12 (40%) and 13 (61.9%) patients in the training and validation cohorts experienced liver regeneration, respectively. The risk stratification model based on the cutoffs of standard residual liver volume spared from at least 20 Gy (SVs20 = 303.4 mL/m2) and alanine aminotransferase (ALT=43 u/L) was able to effectively discriminate the probability of liver regeneration. The model and nomogram of liver regeneration based on SVs20 and ALT showed good prediction performance (AUC=0.759) in the training cohort and performed well (AUC=0.808) in the validation cohort.ConclusionsSVs20 and ALT were optimal predictors of liver regeneration. This model may be beneficial to the constraints of the normal liver outside the radiotherapy-targeted areas.

Appraisal Study of Similarity-Based and Embedding-Based Link Prediction Methods on Graphs

The task of inferring missing links or predicting future ones in a graph based on its current structure is referred to as link prediction. Link prediction methods that are based on pairwise node similarity are well-established approaches in the literature and show good prediction performance in many real-world graphs though they are heuristic. On the other hand, graph embedding approaches learn low-dimensional representation of nodes in graph and are capable of capturing inherent graph features, and thus support the subsequent link prediction task in graph. This appraisal paper studies a selection of methods from both categories on several benchmark (homogeneous) graphs with different properties from various domains. Beyond the intra and inter category comparison of the performances of the methods our aim is also to uncover interesting connections between Graph Neural Network(GNN)-based methods and heuristic ones as a means to alleviate the black-box well-known limitation.

Multispectral Cameras and Machine Learning Integrated into Portable Devices as Clay Prediction Technology

The present work proposed a low-cost portable device as an enabling technology for agriculture using multispectral imaging and machine learning in soil texture. Clay is an important factor for the verification and monitoring of soil use due to its fast reaction to chemical and surface changes. The system developed uses the analysis of reflectance in wavebands for clay prediction. The selection of each wavelength is performed through an LED lamp panel. A NoIR microcamera controlled by a Raspberry Pi device is employed to acquire the image and unfold it in RGB histograms. Results showed a good prediction performance with R2 of 0.96, RMSEC of 3.66% and RMSECV of 16.87%. The high portability allows the equipment to be used in a field providing strategic information related to soil sciences.

Predicting Successes and Failures of Clinical Trials With Outer Product–Based Convolutional Neural Network

Despite several improvements in the drug development pipeline over the past decade, drug failures due to unexpected adverse effects have rapidly increased at all stages of clinical trials. To improve the success rate of clinical trials, it is necessary to identify potential loser drug candidates that may fail at clinical trials. Therefore, we need to develop reliable models for predicting the outcomes of clinical trials of drug candidates, which have the potential to guide the drug discovery process. In this study, we propose an outer product–based convolutional neural network (OPCNN) model which integrates effectively chemical features of drugs and target-based features. The validation results via 10-fold cross-validations on the dataset used for a data-driven approach PrOCTOR proved that our OPCNN model performs quite well in terms of accuracy, F1-score, Matthews correlation coefficient (MCC), precision, recall, area under the curve (AUC) of the receiver operating characteristic, and area under the precision–recall curve (AUPRC). In particular, the proposed OPCNN model showed the best performance in terms of MCC, which is widely used in biomedicine as a performance metric and is a more reliable statistical measure. Through 10-fold cross-validation experiments, the accuracy of the OPCNN model is as high as 0.9758, F1 score is as high as 0.9868, the MCC reaches 0.8451, the precision is as high as 0.9889, the recall is as high as 0.9893, the AUC is as high as 0.9824, and the AUPRC is as high as 0.9979. The results proved that our OPCNN model shows significantly good prediction performance on outcomes of clinical trials and it can be quite helpful in early drug discovery.

Forecasting Air Temperature on Edge Devices with Embedded AI

With the advent of the Smart Agriculture, the joint utilization of Internet of Things (IoT) and Machine Learning (ML) holds the promise to significantly improve agricultural production and sustainability. In this paper, the design of a Neural Network (NN)-based prediction model of a greenhouse’s internal air temperature, to be deployed and run on an edge device with constrained capabilities, is investigated. The model relies on a time series-oriented approach, taking as input variables the past and present values of the air temperature to forecast the future ones. In detail, we evaluate three different NN architecture types—namely, Long Short-Term Memory (LSTM) networks, Recurrent NNs (RNNs) and Artificial NNs (ANNs)—with various values of the sliding window associated with input data. Experimental results show that the three best-performing models have a Root Mean Squared Error (RMSE) value in the range 0.289÷0.402∘C, a Mean Absolute Percentage Error (MAPE) in the range of 0.87÷1.04%, and a coefficient of determination (R2) not smaller than 0.997. The overall best performing model, based on an ANN, has a good prediction performance together with low computational and architectural complexities (evaluated on the basis of the NetScore metric), making its deployment on an edge device feasible.