With the rapid development of Internet technology and artificial intelligence, artificial intelligence technology based on deep learning has created new application prospects in the education industry, promoting the innovation and progress of the education system and facilitating the transformation from traditional teaching mode to intelligent education. In this transformation process, knowledge tracing technology is a key tool for assessing student learning behavior. Knowledge tracing aims to trace and assess students' understanding and mastery of individual knowledge points in real-time by analyzing data from students' previous answers to construct models. With the continuous advancement of deep learning technology in recent years, knowledge tracing has evolved into a mainstream approach for modelling students' mastery of knowledge in educational assessment, predominantly employing Recurrent Neural Networks (RNN). Aiming at the challenges RNN face in handling long-term data dependencies, this paper proposes a hybrid knowledge tracing model that combines Temporal Convolutional Networks (TCN) and Long Short-Term Memory (LSTM). The proposed model is tested and evaluated on three public datasets, ASSISTments2009, ASSISTments2017, and Statics2011. Compared with the existing classical methods, the proposed model shows a significant improvement in the Area Under Curve (AUC) and Accuracy (ACC), which verifies the effectiveness of the proposed method.

This article proposes a method for enhancing vehicle monocular 3-D perception using vehicle reidentification (Re-ID) and collaborative vehicle infrastructure systems (CVIS), aimed at enhancing the perception range and safety of the majority of intelligent connected vehicles currently using cameras. The method initially employs a monocular 3-D perception approach to extract images and rough 3-D information of traffic targets from the vehicle side. Following that, an adaptive compression method called Adaptive-FALSH is introduced, which, combined with vehicle Re-ID technology, enables efficient compression and correlation of vehicle Re-ID features. Ultimately, a perception fusion method dubbed Hamming registration Hamming fusion is proposed, merging monocular 3-D detection results from the vehicle side and high-precision perception results from the roadside. This method quickly extend roadside perception results in real-time to the vehicle side, thereby enhancing the vehicle’s perception range. Experimental results demonstrate that this method efficiently merges 3-D target perception information from both vehicle and road sides using just a few hundred bytes, without the need for high-precision maps and global positioning system assistance. While reducing communication data volume, this method also effectively widens the perception range of intelligent connected vehicles.

The efficient exchange of perception information between vehicles and infrastructure is crucial for implementing vehicle-infrastructure (VI) collaborative intelligent driving. To address the high real-time requirements of VI communication and lack of intelligence and flexibility in VI cooperation, this study proposes an efficient collaborative perception method based on vehicle re-identification for VI collaboration scenarios. The real-time requirements of such scenarios are addressed and a lightweight vehicle re-identification network called ShuffleBNLSH is designed. This network is combined with a hash algorithm to quickly generate ID information for collaborative sensing targets. Based on the state of the VI communication channel, the network can adaptively extract the bit features of the perceived vehicle target, adjust the feature length, and quickly perform feature matching for vehicle target re-identification. To rapidly fuse the VI collaborative perception information combined with the re-identification results and LiDAR 3D perception information from the vehicle and infrastructure, we designed a mini-ICP algorithm that can automatically select feature points and perform point-cloud registration. Experimental results show that the amount of data transmitted by a single target in cooperative sensing can be as small as hundreds of bits during the fusion of sensing targets on the vehicle and infrastructure sides. This reduces the bandwidth requirements for fusing perception targets, accelerates feature transmission and matching, and expands the perception range of VI collaborative autonomous vehicles without GPS information.

Conventional methods such as questionnaires and scales to evaluate learners' learning immersion are influenced by individuals' subjective factors. The non-synchronism between the learning state and after-learning investigation also reduces the accuracy. We propose a new method to evaluate learners' learning immersion based on electroencephalogram (EEG) and support vector machine (SVM). We construct 2 learning scenarios to induce immersive senses: VR video learning for high-level immersion and online English word learning for low-level immersion. To distinguish two immersion levels, students' EEGs are collected. After entering their attention score, relaxation score, the synchronization rate between the 2 scores, high alpha and low beta wave into SVM model, the precision accuracy reaches 87.80%. Taken the classified results and the participants' self-reports together, we find VR devices can create a more immersive environment which improves learners' learning effect. Our findings provide evidence supporting the feasibility of predicting learning immersion levels by physiological recordings.

Summary Due to the increasing attention to online learning, cognitive load has been recently considered as a crucial indicator for judging teenagers' learning state so as to improve both learning and teaching effects. However, some traditional cognitive load measurement methods such as subjective measurement are easily influenced by subjective sensation deviation of subjects. None of them can reflect the cognitive load of learners more precisely. Recently, machine learning–based data modeling has gained more importance in the scenarios of various smart wearables and Internet of things applications. Meanwhile, physiological signals have proven to contribute much to human health assessment. On the basis of the above considerations, this paper proposes a physiological data‐driven model for learners' cognitive load detection under the application of smart wearables. The model consists of four modules: physiological signal acquisition, signal preprocessing, heart rate variability and pulse rate variability feature fusion, and cognitive load classification through an optimized extreme gradient boosting classifier in which hyperparameters are adaptively tuned with sequential model–based optimization. Furthermore, we design an experimental paradigm for signal acquisition in a learning environment, and the experimental results demonstrate that the proposed model for cognitive load detection outperforms conventional approaches that only employ either heart rate variability or pulse rate variability for modeling. We also compare the effects of different feature fusion algorithms combined with different classification algorithms, which demonstrates that the proposed model achieves the highest accuracy of cognitive load detection due to its optimal combination of feature fusion and classification.

Summary The value of image retrieval has become more and more prominent in the era of big data. However, large numbers of images are missed from current method since the image retrieval precision largely depends on the high quality of images. By common methodology, when the quality of images decreases a little, the accuracy of retrieval would decrease significantly. In particular, it is difficult to retrieve noisy images effectively by conventional approach. Yet large number of the noisy images could not be ignored at the age of data explosion. Aiming at the problem above, we proposed noisy image retrieval model based on field of experts (FoE) optimization. High‐quality learning images could be selected by sparse coding, which is based on similarity calculation model, and the multioption filter combination model enhances the power of FoE model. We set up a database containing a large numbers of noisy images. Over this database, adequate groups of experiments are conducted. The verification of the method concluded its effectiveness and superiority.

Summary Analogical reasoning, as a higher cognitive ability, can help children make inferences about a novel situation. It is vital to help children's analogical reasoning development. However, the traditional intervention methods are simple and the effects cannot maintain. Aiming at this problem, the present study was the first to use computer technology especially image composition technique to promote children's analogical reasoning from an interdisciplinary perspective. Specifically, one minimum region entropy based composition model was proposed. On the one hand, sparse coding model and spatial pyramid matching model were used for searching semantically matching images. On the other hand, minimum region entropy model could contribute to composite the candidate region into an ideal position. Furthermore, we set up a database using massive images and adequate experiments based on it to verify the model's effectiveness and robustness. What's more important, we applied the improved image composition to analogical reasoning task. The results showed that the performance of intervention group was obviously better than control group during intervention stages and posttest stage. In general, the present study not only demonstrated the advantages of the improved image composition but also revealed composition's remarkable contribution for getting analogical relationship by children.

Facial expression recognition, which many researchers have put much effort in, is an important portion of affective computing and artificial intelligence. However, human facial expressions change so subtly that recognition accuracy of most traditional approaches largely depend on feature extraction. Meanwhile, deep learning is a hot research topic in the field of machine learning recently, which intends to simulate the organizational structure of human brain's nerve and combine low-level features to form a more abstract level. In this paper, we employ a deep convolutional neural network (CNN) to devise a facial expression recognition system, which is capable to discover deeper feature representation of facial expression to achieve automatic recognition. The proposed system is composed of the Input Module, the Pre-processing Module, the Recognition Module and the Output Module. We introduce both the Japanese Female Facial Expression Database(JAFFE) and the Extended Cohn-Kanade Dataset(CK+) to simulate and evaluate the recognition performance under the influence of different factors (network structure, learning rate and pre-processing). We also introduce a K-nearest neighbor (KNN) algorithm compared with CNN to make the results more convincing. The accuracy performance of the proposed system reaches 76.7442% and 80.303% in the JAFFE and CK+, respectively, which demonstrates feasibility and effectiveness of our system.

Target localization and tracking in the Internet of Things (IoT) environment have been paid more and more attention recently. The knowledge and information generated from wireless sensor nodes of the IoT make huge contributions to localization and tracking of targets with high mobility. This paper presents a square-root unscented Kalman filtering (SR-UKF) based localization and tracking algorithm for mobile target in an IoT environment. First, a localization initialization model is proposed for an IoT scenario. Then, according to information of neighboring sensor nodes, we employ the SR-UKF idea for the further localization and tracking of the target. Simulation results demonstrate that the proposed algorithm achieves lower localization and tracking error under the same computational complexity, compared with some conventional extended Kalman filtering (EKF) or UKF based methods. The proposed algorithm is of great significance in the field of IoT information processing.