Identifying drug-target interactions (DTI) is a fundamental yet costly step in drug discovery, motivating the development of accurate and efficient computational prediction methods. In this paper, we propose FKAN-a, a DTI prediction framework that integrates Fourier-enhanced Kolmogorov-Arnold networks (KAN) with attention mechanisms under a contrastive learning paradigm. Drug molecules and protein sequences are preprocessed and encoded using pretrained representations to capture interaction-relevant features. The resulting embeddings are further transformed through KAN with learnable Fourier bases to model complex nonlinear relationships. A cross-modality attention module is introduced to enhance the modeling of fine-grained drug-protein associations. Experiments conducted on three public benchmark datasets demonstrate that FKAN-a consistently outperforms representative state-of-the-art methods in terms of prediction performance and computational efficiency. These results indicate that the proposed framework provides an effective solution for DTI prediction and practical candidate prioritization in drug discovery.

Arrhythmia is a prevalent cardiovascular disease, which has garnered widespread attention due to its age‐related increases in mortality rates. In the analysis of arrhythmia, the electrocardiogram (ECG) plays an important role. Arrhythmia classification often suffers from a significant data imbalance issue due to the limited availability of data for certain arrhythmia categories. This imbalance problem significantly affects the classification performance of the model. To address this challenge, data augmentation emerges as a viable solution, aiming to neutralize the adverse effects of imbalanced datasets on the model. To this end, this paper proposes a novel Multimodality Data Augmentation Network (MM‐DANet) for arrhythmia classification. The MM‐DANet consists of two modules: the multimodality data matching‐based data augmentation module and the multimodality feature encoding module. In the multimodality data matching‐based data augmentation module, we expand the underrepresented arrhythmia categories to match the size of the largest category. Subsequently, the multimodality feature encoding module employs convolutional neural networks (CNN) to extract the modality‐specific features from both signals and images and concatenate them for efficient and accurate classification. The MM‐DANet was evaluated on the MIT‐BIH Arrhythmia Database and achieving an accuracy of 98.83%, along with an average specificity of 98.87%, average sensitivity of 92.92%, average precision of 91.05%, and average F 1_score of 91.96%. Furthermore, its performance was also assessed on the St. Petersburg INCART arrhythmia database and the MIT‐BIH supraventricular arrhythmia database, yielding AUC values of 81.98% and 90.93%, respectively. These outstanding results not only underscore the effectiveness of MM‐DANet but also indicate its potential for facilitating reliable automated analysis of arrhythmias.

Genome-wide association studies have shown that common genetic variants associated with complex diseases are mostly located in non-coding regions, which may not be causal. In addition, the limited number of validated non-coding functional variants makes it difficult to develop an effective supervised learning model. Therefore, improving the accuracy of predicting non-coding causal variants has become critical. This study aims to build a transfer learning-based machine learning method for predicting regulatory variants to overcome the problem of limited sample size. This paper presents a supervised learning method transfer support vector machine (TSVM) for massively parallel reporter assays (MPRA) validated regulatory variants prediction. First, uses a convolutional neural network to extract features with transfer learning. Second, the extracted features are selected by random forest method. Third, the selected features are used to train support vector machine for classification. We performed scale sensitivity experiments on the MPRA dataset and validated the effectiveness of transfer learning. The model achieves the Mcc of 0.326 and the AUC of 0.720, which are higher than the state-of-the-art method.

The function labeling of enzymes has a wide range of application value in the medical field, industrial biology and other fields. Scientists define enzyme categories by enzyme commission (EC) numbers. At present, although there are some tools for enzyme function prediction, their effects have not reached the application level. To improve the precision of enzyme function prediction, we propose a parallel convolutional contrastive learning (PCCL) method to predict enzyme functions. First, we use the advanced protein language model ESM-2 to preprocess the protein sequences. Second, PCCL combines convolutional neural networks (CNNs) and contrastive learning to improve the prediction precision of multifunctional enzymes. Contrastive learning can make the model better deal with the problem of class imbalance. Finally, the deep learning framework is mainly composed of three parallel CNNs for fully extracting sample features. we compare PCCL with state-of-art enzyme function prediction methods based on three evaluation metrics. The performance of our model improves on both two test sets. Especially on the smaller test set, PCCL improves the AUC by 2.57%.

RNA-binding proteins are important for the process of cell life activities. High-throughput technique experimental method to discover RNA-protein binding sites is time-consuming and expensive. Deep learning is an effective theory for predicting RNA-protein binding sites. Using weighted voting method to integrate multiple basic classifier models can improve model performance. Thus, in our study, we propose a weighted voting deep learning model (WVDL), which uses weighted voting method to combine convolutional neural network (CNN), long short term memory network (LSTM) and residual network (ResNet). First, the final forecast result of WVDL outperforms the basic classifier models and other ensemble strategies. Second, WVDL can extract more effective features by using weighted voting to find the best weighted combination. And, the CNN model also can draw the predicted motif pictures. Third, WVDL gets a competitive experiment result on public RBP-24 datasets comparing with other state-of-the-art methods. The source code of our proposed WVDL can be found in https://github.com/biomg/WVDL.

Computational prediction of the RBP bound sites using features learned from existing annotation knowledge is an effective method because high-throughput experiments are complex, expensive and time-consuming. Many methods have been proposed to predict RNA-protein binding sites. However, the partial information of RNA sequence is not fully used. In this study, we propose multiple convolutional neural networks (MCNN) method, which predicts RNA-protein binding sites by integrating multiple convolutional neural networks constructed by RNA sequence information extracted from windows with different lengths. First, MCNN trains multiple CNNs base on RNA sequences extracted by different window lengths. Second, MCNN can extract more binding patterns of RBPs by combining these trained multiple CNNs previously. Third, MCNN only uses RNA base sequence information for RNA-protein binding sites prediction, which extracts sequence binding features and predicts the result with same architecture. This avoids the information loss of feature extraction step. Our proposed MCNN demonstrates a competitive performance comparing with other methods on a large-scale dataset derived from CLIP-seq, which is an effective method for RNA-protein binding sites prediction. The source code of our proposed MCNN method can be found in https://github.com/biomg/MCNN.

Objective: Green computing meets the needs of a low-carbon society and it is an important aspect of promoting social sustainable development and technological progress. In the investigation, green computing for resource management and allocation issues is only discussed. Therefore, in the context of the 5G communication network, the investigation of the data classification and resource optimization of the Internet of Things are conducted. Method: The virtualization architecture of the heterogeneous wireless network resource based on 5G technology is designed. The related investigation is conducted based on 5G network and Internet of Things technology. Under the traditional method, the transfer learning is introduced to improve the AdaBoost (Adaptive Boosting) algorithm to classify the data. The investigated complete resource reuse method is used to optimize resources. A method that a sub-channel can be reused by a cellular link and any number of D2D links at the same time is proposed to conduct resource optimization investigation. Results: The investigation indicates that the classification accuracy of the algorithm is excellent for the data classification of the Internet of Things and has different advantages in various aspects compared with other algorithms. The designed algorithm can find a larger set of resource reuse and have a significant increase in spectrum utilization efficiency. Conclusion: The investigation can contribute to the boom in the Internet of Things in terms of data classification and resource optimization based on 5G.

The brain-computer interface (BCI) connects the brain and the external world through an information transmission channel by interpreting the physiological information of the brain during thinking activities. The effective classification of electroencephalogram (EEG) signals is the key to improving the performance of the system. To improve the classification accuracy of EEG signals in the BCI system, the transfer learning algorithm and the improved Common Spatial Pattern (CSP) algorithm are combined to construct a data classification model. Finally, the effectiveness of the proposed algorithm is verified. The results show that in actual and imagined movements, the accuracy of the left- and right-hand movements at different speeds is higher than when the speeds are the same. The proposed Adaptive Composite Common Spatial Pattern (ACCSP) and Self Adaptive Common Spatial Pattern (SACSP) algorithms have good classification effects on 5 subjects, with an average classification accuracy rate of 83.58 percent, which is an increase of 6.96 percent compared with traditional algorithms. When the training sample size is 10, the classification accuracy of the ACCSP algorithm is higher than that of the traditional CSP algorithm. The improved CSP algorithm combined with transfer learning embodies a good classification effect in both ACCSP and SACSP. Especially, the performance of SACSP mode is better. Combining the improved CSP algorithm proposed with the CSP-based transfer learning algorithm can improve the classification accuracy of the BCI classifier.

To investigate the diversified technologies in Internet of Vehicles (IoV) under intelligent edge computing, artificial intelligence, intelligent edge computing, and IoV are combined. Also, it proposes an IoV model for intelligent edge computing task offloading and migration under the SDVN (Software Defined Vehicular Networks) architecture, that is, the JDE-VCO (Joint Delay and Energy-Vehicle Computational task Offloading) optimization. And the simulation is performed. The results show that in the analysis of the impact of different offloading strategies on the IoV, it is found that the JDE-VCO algorithm is superior to other schemes in terms of transmission delay and total offloading energy consumption. In the analysis of the impact of the task unloading of the IoV, the JDE-VCO algorithm is less than RTO (Random Tasks Offloading) and UTO (Uniform Tasks Offloading) algorithm schemes in terms of the number of tasks per unit time, and the average task completion time for the same amount of uploaded data. In the analysis of the packet loss ratio and transmission delay, it can be found that the packet loss ratio and transmission delay of the JDE-VCO algorithm are less than the RTO and UTO algorithms. Moreover, the packet loss ratio of the JDE-VCO algorithm is about 0.1, and the transmission delay is stable at 0.2s, which has obvious advantages. Therefore, through research, the IoV model of task offloading and migration built by intelligent edge computing can significantly improve the load sharing rate, offloading efficiency, packet loss ratio, and transmission delay when the IoV is processing tasks and uploading data. It provides experimental basis for the improvement of the IoV system.

To explore the electric vehicle networks in smart cities through big data analysis technology, this study utilizes K-means and fuzzy theory in big data analysis technology to construct an objective function-based fuzzy mean clustering algorithm theory (FCM). Then, the FCM algorithm is improved, and the electric vehicle network is simulated. The results show that in the analysis of network data transmission performance, when the probability of successful propagation is 100% and the λ value is between 0.01-0.05, it is closest to the actual result, and the data delay is the smallest. In the analysis of the route guidance effects, when facing congested road sections, the route guidance strategy of this study can restrain the spread of congestion effectively and achieve timely evacuation of traffic congestion. In the further analysis of the impact of different factors on traffic conditions, under route guidance, with the increase in market penetration rate (MPR) of devices, following rate (FR) of vehicles, and congestion level (CL), the improvement of the induction strategy becomes clearer, and greater economic benefits are achieved. This study has found that utilizing big data analysis technology to improve the electric vehicle transportation networks can reduce the network data transmission performance delay significantly and change the path to suppress the spread of congestion effectively, which has provided experimental references for the development of electric vehicle transportation networks.

Smart cities fully utilize the new generation of Internet of Things (IoT) technology in the process of urban informatization to optimize the urban management and service. However, in the IoT system, while information exchange and communication, wireless sensor network devices may not be able to resist all forms of attacks, which may lead to security issues such as user data disclosure. Aiming at the information security risks in smart city, the typical technologies in IoT is analyzed from the perspective of IoT perception layer and provides corresponding security solutions for the existing security threats. Regarding the communication security, the emerging wireless technology, long range (LoRa), is discussed, and the performance of wireless communication protocol is analyzed through simulation experiments, to verify that the IoT technology based on LoRa communication technology can improve the security of the system in the construction of smart city. The results show that REBEB, a new backoff algorithm, is similar to the binary exponential backoff algorithm in terms of throughput performance. REBEB focuses more on fairness, which is up to 0.985, and to a certain extent, its security is significantly improved. The fairness of REBEB algorithm is more than 0.4 in different nodes and competing windows, and the fairness of the system is better when the number of nodes is small. To sum up, the IoT system based on LoRa communication can effectively improve the security performance of the system in the construction of smart city and avoid the security threats in the IoT signal transmission.

Emotional cognitive ability is a key technical indicator to measure the friendliness of interaction. Therefore, this research aims to explore robots with human emotion cognitively. By discussing the prospects of 5G technology and cognitive robots, the main direction of the study is cognitive robots. For the emotional cognitive robots, the analysis logic similar to humans is difficult to imitate; the information processing levels of robots are divided into three levels in this study: cognitive algorithm, feature extraction, and information collection by comparing human information processing levels. In addition, a multi-scale rectangular direction gradient histogram is used for facial expression recognition, and robust principal component analysis algorithm is used for facial expression recognition. In the pictures where humans intuitively feel smiles in sad emotions, the proportion of emotions obtained by the method in this study are as follows: calmness accounted for 0%, sadness accounted for 15.78%, fear accounted for 0%, happiness accounted for 76.53%, disgust accounted for 7.69%, anger accounted for 0%, and astonishment accounted for 0%. In the recognition of micro-expressions, humans intuitively feel negative emotions such as surprise and fear, and the proportion of emotions obtained by the method adopted in this study are as follows: calmness accounted for 32.34%, sadness accounted for 34.07%, fear accounted for 6.79%, happiness accounted for 0%, disgust accounted for 0%, anger accounted for 13.91%, and astonishment accounted for 15.89%. Therefore, the algorithm explored in this study can realize accuracy in cognition of emotions. From the preceding research results, it can be seen that the research method in this study can intuitively reflect the proportion of human expressions, and the recognition methods based on facial expressions and micro-expressions have good recognition effects, which is in line with human intuitive experience.

With increasing amounts of information, the image information received by people also increases exponentially. To perform fine-grained categorization and recognition of images and visual calculations, this study combines the Visual Geometry Group Network 16 model of convolutional neural networks and the vision attention mechanism to build a multi-level fine-grained image feature categorization model. Finally, the TensorFlow platform is utilized to simulate the fine-grained image classification model based on the visual attention mechanism. The results show that in terms of accuracy and required training time, the fine-grained image categorization effect of the multi-level feature categorization model constructed by this study is optimal, with an accuracy rate of 85.3% and a minimum training time of 108 s. In the similarity effect analysis, it is found that the chi-square distance between Log Gabor features and the degree of image distortion show a strong positive correlation; in addition, the validity of this measure is verified. Therefore, through the research in this study, it is found that the constructed fine-grained image categorization model has higher accuracy in image recognition categorization, shorter training time, and significantly better performance in similar feature effects, which provides an experimental reference for the visual computing of fine-grained images in the future.

With the continuous expansion of the application scope of computer network technology, various malicious attacks that exist in the Internet range have caused serious harm to computer users and network resources. This paper attempts to apply artificial intelligence (AI) to computer network technology and research on the application of AI in computing network technology. Designing an intrusion detection model based on improved back propagation (BP) neural network. By studying the attack principle, analyzing the characteristics of the attack method, extracting feature data, establishing feature sets, and using the agent technology as the supporting technology, the simulation experiment is used to prove the improvement effect of the system in terms of false alarm rate, convergence speed, and false negative rate, the rate reached 86.7%. The results show that this fast algorithm reduces the training time of the network, reduces the network size, improves the classification performance, and improves the intrusion detection rate.

This paper proposed a high-performance image retrieval framework, which combines the improved feature extraction algorithm SIFT (Scale Invariant Feature Transform), improved feature matching, improved feature coding Fisher and improved Gaussian Mixture Model (GMM) for image retrieval. Aiming at the problem of slow convergence of traditional GMM algorithm, an improved GMM is proposed. This algorithm initializes the GMM by using on-line [Formula: see text]-means clustering method, which improves the convergence speed of the algorithm. At the same time, when the model is updated, the storage space is saved through the improvement of the criteria for matching rules and generating new Gaussian distributions. Aiming at the problem that the dimension of SIFT (Scale Invariant Feature Transform) algorithm is too high, the matching speed is too slow and the matching rate is low, an improved SIFT algorithm is proposed, which preserves the advantages of SIFT algorithm in fuzzy, compression, rotation and scaling invariance advantages, and improves the matching speed, the correct match rate is increased by an average of 40% to 55%. Experiments on a recently released VOC 2012 database and a database of 20 category objects containing 230,800 images showed that the framework had high precision and recall rates and less query time. Compared with the standard image retrieval framework, the improved image retrieval framework can detect the moving target quickly and effectively and has better robustness.

Human action recognition based on depth video sequence is an important research direction in the field of computer vision. The present study proposed a classification framework based on hierarchical multi-view to resolve depth video sequence-based action recognition. Herein, considering the distinguishing feature of 3D human action space, we project the 3D human action image to three coordinate planes, so that the 3D depth image is converted to three 2D images, and then feed them to three subnets, respectively. With the increase of the number of layers, the representations of subnets are hierarchically fused to be the inputs of next layers. The final representations of the depth video sequence are fed into a single layer perceptron, and the final result is decided by the time accumulated through the output of the perceptron. We compare with other methods on two publicly available datasets, and we also verify the proposed method through the human action database acquired by our Kinect system. Our experimental results demonstrate that our model has high computational efficiency and achieves the performance of state-of-the-art method.

Fatigue driving is bringing more and more serious harm, but there are various reasons for fatigue driving, it is still difficult to test the driver’s fatigue. This paper defines a method to test driver’s fatigue based on the EEG, and different from other researches into fatigue driving, this paper mainly takes the fatigue features of EEG signals in fatigue state and uses wavelet entropy as the feature extraction method to analyze the features of wavelet entropy and spectral entropy features as well as the classification accuracy under the same classifier. The SVM is used to show the classifier’s results. The accuracy of the driver fatigue state monitoring using the wavelet entropy is 90.7%, which is higher than the use of spectral entropy as the characteristic accuracy rate of 81.3%. The results show that the frequency characteristics of EEG can be well applied to driving fatigue testing, but different frequency feature calculation methods will affect the classification accuracy.