Variable Neural Network Research Articles

Federated Learning‐Based Mobile Traffic Prediction in Satellite‐Terrestrial Integrated Networks

ABSTRACTIntroductionWith the development and integration of satellite and terrestrial networks, mobile traffic prediction has become more important than before, which is the basis for service provision and resource scheduling when supporting various vertical applications. However, existing traffic prediction methods, especially deep learning‐based methods, require massive data for model training. Due to data privacy concerns, mobile traffic data are not easily shared among different parties, making it difficult to obtain a precise prediction model.MethodsTo mitigate the data leakage risk, a federated learning framework is proposed in this study for mobile traffic prediction in satellite‐terrestrial integrated networks to achieve a tradeoff between data privacy and prediction accuracy. In the proposed framework, local models are trained in base stations on the ground, and a global model is aggregated in the satellite edge server in space.ResultsA deep learning‐based prediction model with an adaptive graph convolutional network (AGCN) and long short‐term memory (LSTM) modules is proposed and validated in numerical experiments, which achieves the lowest prediction error with a real‐world traffic dataset when compared with other graph neural network (GNN) variants in the federated learning setting.ConclusionNumerical experiments with a real‐world mobile traffic dataset demonstrate the effectiveness of the proposed approach, which outperforms other GNN variants with lower prediction errors.

Decision model of public opinion risk in campus social network based on hybrid dynamic deletion and shortest path algorithm

Aiming at the problem that traditional network public opinion monitoring and searching are inefficient and can easily cause resource waste, the study firstly, through the dynamic deletion-shortest path algorithm to classify network text, and on this basis, innovatively constructs a text sentiment classification model based on the variant of convolutional neural network and recurrent neural network, and secondly, uses attention mechanism to classify the model. improvement of the classification model by using the attention mechanism. The research results show that the average precision rate, recall rate, and F-value of the dynamic deletion-shortest path algorithm are 97.30%, 79.55%, and 87.53%, and the classification speed is 397 KB/s, which is better than the traditional shortest path algorithm. In the classification effect measurement of long text, the accuracy and F-value of the recurrent neural network variant model are above 84%, and the accuracy of the text sentiment classification model with the introduction of the attention mechanism is improved by 3.89% compared to the pre-improvement period. In summary, the dynamic deletion-shortest path algorithm proposed in the study and the sentiment classification model with the introduction of the attention mechanism have superior performance and can provide certain application value for campus social network opinion risk decision-making.

Graph neural network-based subgraph analysis for predicting adverse drug events

PurposeAdverse drug events (ADEs) are a significant global public health concern, and they have resulted in high rates of hospital admissions, morbidity, and mortality. Prior to the use of machine learning and deep learning methods, ADEs may not become well recognized until long after a drug has been approved and is widely used, which poses a significant challenge for ensuring patient safety. Consequently, there is a need to develop computational approaches for earlier identification of ADEs not detected during pre-registration clinical trials. MethodsThis paper presents a state-of-the-art network-based approach that models patients as subgraphs composed of nodes of International Classification of Diseases (ICD) codes and directed edges illustrating disease progression. Four Graph Neural Network (GNN) variants were employed to make sub-graph level predictions that answer three Research Questions (RQ): 1) whether ADE(s) would occur given a patient's prior diagnoses history, 2) when an ADE would occur, and 3) which ADE would occur. The first and second RQs were addressed using a binary classification approach. The third RQ was addressed using a multi-label classification model. ResultsThe proposed network-based approach demonstrated superior performance in predicting ADEs, with the GraphSage model exhibiting the highest accuracy for both RQ 1 (0.8863) and RQ 3 (0.9367), while the Graph Attention Networks (GAT) model was found to perform best for RQ 2 (0.8769). Furthermore, an analysis segmented by ADE classification revealed that while RQs 1 and 3 exhibited minimal variance across different ADE categories, a distinct advantage was observed for categories B, C, and E in the context of RQ 2 when applying this sub-graph method. ConclusionThe network-based approach demonstrates the potential of GNNs in supporting the early detection and prevention of ADEs. Accurately predicting ADEs could enable healthcare professionals to make informed clinical decisions, take preventive measures and adjust medication regimens before serious adverse events occur. The proposed prediction method could also lead to optimized usage of healthcare resources by preventing hospital admissions and reducing the overall burden of adverse drug events on the healthcare systems.

Comparative analysis and application of soft sensor models in domestic wastewater treatment for advancing sustainability

ABSTRACT This study focuses on the development and evaluation of soft sensor models for predicting NH3-N values in a wastewater treatment process. The study compares the performance of linear regression (LR), neural networks (NN) and random forest regression (RFR) models. The proposed methodology involves optimizing the sequencing batch reactor process using artificial intelligence and an automatic control system. Real-time NH3-N values are obtained by inputting data from electronic conductivity and temperature sensors into the prediction models. Once the predicted NH3-N value falls below the effluent standard, the cycle ends, improving energy efficiency and sustainability by cutting down the agitator and aerator. The research results demonstrate that the RNN-based NH3-N soft sensor built in this study exhibits the best performance, which is promising for wastewater treatment process optimization and evaluation. The results show that sensor model NNR[0.5Y]H exhibits exceptional performance, utilizing recurrent neural network with 5-step input delays. Sensor NNR[0.5Y]H exhibits an R2 of 0.921, an RMSE of 6.110, and an MAE of 4.558. Based on the findings, recurrent neural network (RNN) variants emerge as the most effective modeling technique due to their ability to capture temporal dependencies and handle variable-length sequences. This study provides satisfied performance results for the NNR[0.5Y]H soft sensor model in NH3-N monitoring and process optimization in wastewater treatment, highlighting the effectiveness of recurrent neural networks and their contribution to improving interpretability, accuracy, and adaptability of soft sensor models.

Enhanced Malaysian License Plate Recognition System Using an Improved YOLOv2 Model

License Plate Recognition (LPR) has gained popularity among researchers due to its wide range of applications, including law enforcement, monitoring, and toll gate systems. However, existing LPR systems still require improvements to achieve optimum accuracy and speed. The advancements in Convolutional Neural Network (CNN) variants offer potential solutions for these challenges. This primary aim of this system is to ensure accurate and efficient recognition of the vehicle plate characters using CNN techniques. This research utilizes two CNN network architectures for deep object detection to address the Malaysian License Plate Recognition (MLPR) task. The first network is designed to detect the license plate, while the second is responsible for recognizing the characters on the plate. Both networks are cascaded from the architecture of two-stage YOLOv2, providing promising speed and accuracy. The MLPR system achieved an accuracy of 98.75% and a processing speed of 0.0104 seconds, using a total of 2,200 license plate images. In conclusion, the system adapted from deep object detection techniques presents a promising solution for the MLPR problem, based on the achieved accuracy and speed.

Handrim Reaction Force and Moment Assessment Using a Minimal IMU Configuration and Non-Linear Modeling Approach during Manual Wheelchair Propulsion.

Manual wheelchair propulsion represents a repetitive and constraining task, which leads mainly to the development of joint injury in spinal cord-injured people. One of the main reasons is the load sustained by the shoulder joint during the propulsion cycle. Moreover, the load at the shoulder joint is highly correlated with the force and moment acting at the handrim level. The main objective of this study is related to the estimation of handrim reactions forces and moments during wheelchair propulsion using only a single inertial measurement unit per hand. Two approaches are proposed here: Firstly, a method of identification of a non-linear transfer function based on the Hammerstein-Wiener (HW) modeling approach was used. The latter represents a typical multi-input single output in a system engineering modeling approach. Secondly, a specific variant of recurrent neural network called BiLSTM is proposed to predict the time-series data of force and moments at the handrim level. Eleven subjects participated in this study in a linear propulsion protocol, while the forces and moments were measured by a dynamic platform. The two input signals were the linear acceleration as well the angular velocity of the wrist joint. The horizontal, vertical and sagittal moments were estimated by the two approaches. The mean average error (MAE) shows a value of 6.10 N and 4.30 N for the horizontal force for BiLSTM and HW, respectively. The results for the vertical direction show a MAE of 5.91 N and 7.59 N for BiLSTM and HW, respectively. Finally, the MAE for the sagittal moment varies from 0.96 Nm (BiLSTM) to 1.09 Nm for the HW model. The approaches seem similar with respect to the MAE and can be considered accurate knowing that the order of magnitude of the uncertainties of the dynamic platform was reported to be 2.2 N for the horizontal and vertical forces and 2.24 Nm for the sagittal moments. However, it should be noted that HW necessitates the knowledge of the average force and patterns of each subject, whereas the BiLSTM method do not involve the average patterns, which shows its superiority for time-series data prediction. The results provided in this study show the possibility of measuring dynamic forces acting at the handrim level during wheelchair manual propulsion in ecological environments.

A multi-modal Parkinson’s disease diagnosis system from EEG signals and online handwritten tasks using grey wolf optimization based deep learning model

Parkinson’s disease (PD) is a neurodegenerative disorder characterized by the gradual deterioration of motor function, affecting speech, writing, muscle control, and mobility. The existing studies have not utilized both the electroencephalography (EEG) signals and online handwritten tasks together to diagnose PD. The studies have also not explored the EEG signals collected from specific brain regions like the substansia niagra (SN) and ventral tegmental area (VTA), crucial for dopamine production linked to PD. This article proposes a multi-modal PD diagnosis system from EEG signals (collected from SN and VTA regions of the brain), collected during performing online handwritten tasks, using grey wolf optimization (GWO) algorithm. Mel-frequency cepstral coefficients (MFCC) features have been generated from the EEG signals and optimized by the GWO algorithm. The classification (diagnosis) experiments on the optimal number of feature values, obtained from GWO algorithm, have been carried out using bidirectional long short-term memory (BLSTM) variant of recurrent neural network (RNN). The classification experiments have also been conducted using support vector machine (SVM), bagged random forest (BRF), and long short-term memory (LSTM) variant of RNN classifier to have a performance comparison with the proposed method. The effectiveness of the introduced PD diagnosis system has been analyzed on a self-generated dataset named EEG signal based on online handwriting (ESOH). A maximum classification accuracy of 99.30% has been achieved from the proposed PD diagnosis system. The experimental outcomes illustrate that the introduced PD diagnosis system outperforms the state-of-the-art PD diagnosis systems relying on the EEG signals for diagnosing the PD.

ConvNext as a Basis for Interpretability in Coffee Leaf Rust Classification

The increasing complexity of deep learning models can make it difficult to interpret and fit models beyond a purely accuracy-focused evaluation. This is where interpretable and [ eXplainable Artificial Intelligence (XAI)explainable Artificial Intelligence (AI) come into play to facilitate an understanding of the inner workings of models. Consequently, alternatives have emerged, such as [class activation mapping (CAM) techniquesClass Activation Maps (CAM) aimed at identifying regions of importance for an image classification model. However, the behavior of such models can be highly dependent on the type of architecture and the different variants of convolutional neural networks. Accordingly, this paper evaluates three Convolutional Neural Network (CNN) architectures (VGG16, ResNet50, ConvNext-T) against seven CAM models (GradCAM, XGradCAM, HiResCAM, LayerCAM, GradCAM++, GradCAMElementWise, and EigenCAM), indicating that the CAM maps obtained with ConvNext models show less variability among them, i.e., they are less dependent on the selected CAM approach. This study was performed on an image dataset for the classification of coffee leaf rust and evaluated using the RemOve And Debias (ROAD) metric.

Dynamic Physics-Guided Deep Learning for Long-Term Production Forecasting in Unconventional Reservoirs

Summary Neural network predictive models are popular for production forecasting in unconventional reservoirs due to their ability to learn complex relationships between well properties and production responses from extensive field data. The intricate flow behavior in hydraulically fractured unconventional reservoirs, which remains poorly understood, makes these statistical models particularly useful. Various neural network variants have been developed for production prediction in these reservoirs, each offering predictive capability of varying levels of granularity, accuracy, and robustness against noisy and incomplete data. Neural network predictive models that integrate physical principles are especially useful for subsurface systems, as they provide predictions that adhere to physical laws. This work introduces a new dynamic physics-guided deep learning (DPGDL) model that incorporates physical functions into neural networks and employs residual learning to compensate for the imperfect description of the physics, under variable data support. The new formulation allows for dynamic residual correction, avoids unintended bias due to less-than-ideal input data, and provides robust long-term predictions. The DPGDL model improves upon a static formulation by utilizing a masked loss function to enable learning from wells with varying production lengths and by improving the results when partially-observed timesteps are present. In addition, a sequence-to-sequence residual model has been developed to correct additional biases in the long-term predictions from the physics-constrained neural networks. Several synthetic data sets with increasing complexity as well as a field data set from the Bakken are used to demonstrate the performance of the new DPGDL model.

Photonic deep residual time-delay reservoir computing

Time-delay reservoir computing (TDRC) represents a simplified variant of recurrent neural networks, employing a nonlinear node with a feedback mechanism to construct virtual nodes. The capabilities of TDRC can be enhanced by transitioning to a deep architecture. In this work, we propose a novel photonic deep residual TDRC (DR-TDRC) with augmented capabilities. The additional time delay added to the residual structure enables DR-TDRC superior to traditional deep structures across various benchmark tasks, especially in memory capability and almost an order of magnitude improvement in nonlinear channel equalization. Additionally, a specifically designed clipping algorithm is utilized to counteract the damage of redundant layers in deep structures, enabling the extension of the deep TDRC to dozens rather than just a few layers, with higher performance. We experimentally demonstrate the proof-of-concept with a 4-layer DR-TDRC containing 960 interrelated neurons (240 neurons per layer), based on four injection-locked distributed feedback lasers. We confirm the potential for scalable deep RC with elevated performance. Our results provide a feasible approach for expanding deep photonic computing to satisfy the boosting demand for artificial intelligence.

The performance of encoder–decoder neural networks for leak detection in water distribution networks

ABSTRACT This work outlines the performance of three variants of deep neural networks for leak detection in water distribution networks, namely – autoencoders (AEs), variational autoencoders (VAEs), and long short-term memory autoencoders (LSTM-AEs). The multivariate pressure signals reconstructed from these models are analysed for leakage identification. The leak onset time is estimated using a fast approximation sliding window technique, which computes statistical discrepancies in prediction errors. The performance of all three variants is validated using the widely studied L-Town benchmark network. Furthermore, their feasibility for real-world application is studied by applying them to a real-world case study representing the data availability and network design often found in smaller- and medium-sized utilities in Norway. The results for the benchmark network showed that AE and LSTM-AE showed comparable detection performance for abrupt leaks with VAE performing the least. For incipient leaks, the LSTM-AE showed better detection performance with few false-positives. For the real-world dataset, the performance was significantly lower due to the quantity and quality of data available, and the contradiction of inherent requirements of data-driven models. In addition, the analysis revealed that the positioning of pressure sensors in the network is critical for the leak detection performance of these models.

A New Fingerprint and Graph Hybrid Neural Network for Predicting Molecular Properties.

Machine learning plays a role in accelerating drug discovery, and the design of effective machine learning models is crucial for accurately predicting molecular properties. Characterizing molecules typically involves the use of molecular fingerprints and molecular graphs. These are input into a multilayer perceptron (MLP) and variants of graph neural networks, such as graph attention networks (GATs). Due to the diverse types and large dimension of fingerprints, models may contain many features that are relatively irrelevant or redundant; meanwhile, although the GAT excels in handling heterogeneous graph tasks, it lacks the ability to extract collaborative information from neighboring nodes, which is crucial in scenarios where it cannot capture the joint influence of adjacent groups on atoms. To overcome these challenges, we introduce a hybrid model, combining improved GAT and MLP. In GAT, the recurrent neural network is employed to capture collaborative information. To address the dimensionality issue, we propose a feature selection algorithm, which is based on the principle of maximizing relevance while minimizing redundancy. Through experiments on 13 public data sets and 14 breast cell lines, our model demonstrates superior performance compared to state-of-the-art deep learning and traditional machine learning algorithms. Additionally, a series of ablation experiments were conducted to demonstrate the advantages of our improved version, as well as its antinoise capability and interpretability. These results indicate that our model holds promising prospects for practical applications.

Cloud-based machine learning algorithms for anomalies detection

Gradient boosting machines harnesses the inherent capabilities of decision trees and meticulously corrects their errors in a sequential fashion, culminating in remarkably precise predictions. Word2Vec, a prominent word embedding technique, occupies a pivotal role in natural language processing (NLP) tasks. Its proficiency lies in capturing intricate semantic relationships among words, thereby facilitating applications such as sentiment analysis, document classification, and machine translation to discern subtle nuances present in textual data. Bayesian networks introduce probabilistic modeling capabilities, predominantly in contexts marked by uncertainty. Their versatile applications encompass risk assessment, fault diagnosis, and recommendation systems. Gated recurrent units (GRU), a variant of recurrent neural networks, emerges as a formidable asset in modeling sequential data. Both training and testing are crucial to the success of an intrusion detection system (IDS). During the training phase, several models are created, each of which can recognize typical from anomalous patterns within a given dataset. To acquire passwords and credit card details, "phishing" usually entails impersonating a trusted company. Predictions of student performance on academic tasks are improved by hyper parameter optimization of the gradient boosting regression tree using the grid search approach.

A lightweight deep learning-based android malware detection framework

Android, as the most prevalent mobile operating system (OS) in recent years, has been widely applied in various cell phones, tablets, and embedded devices, greatly facilitating people’s lives. However, Android malware is also emerging, which significantly endangers people’s information security. The current Android malware detection systems are often suffering from cumbersome structures and massive computational resources, which seriously limits their direct deployment on mobile devices. We design a lightweight Android malware detection system named MCADS, which consists of a two-layer structure. At the first layer, we use an augmented Multilayer Perceptron (MLP) for the preliminary analysis of malware. At the second layer, we propose a new lightweight variant of Convolutional Neural Network (CNN) to further analyze the Apps that cannot be accurately identified in the first layer. Finally, a careful experimental investigation has been conducted. The proposed method achieves an accuracy of 98.12%, surpassing that of the compared methods. The promising results demonstrate the potential of MCADS as a practical adjunctive solution for detecting malware, complementing existing complex detection methods.

Automatic classification of multi-carrier modulation signal using STFT spectrogram and deep CNN

In the realm of communication systems, categorizing Multi-Carrier Modulation (MCM) signals without cooperative communication poses a significant technical challenge. In this paper, we introduce a novel approach for accurately categorizing five distinct MCM signals, including Orthogonal Frequency Division Multiplexing (OFDM), Filter Bank Multicarrier (FBMC), Filtered Orthogonal Frequency Division Multiplexing (FOFDM), Windowed Orthogonal Frequency Division Multiplexing (WOLA), and Universal Filtered Multicarrier (UFMC). Each signal is considered with two types of subcarrier waveforms, Quadrature Amplitude Modulation 16 (QAM16) and Quadrature Amplitude Modulation 64 (QAM64), resulting in a total of 10 unique MCM signals for classification. Our proposed methodology leverages Short-Time Fourier Transform (STFT) spectrograms for feature extraction at the frontend, while at the backend, we employ three variants of Convolutional Neural Network (CNN) models; CNN, CNN with Dropout (CNN_d), CNN with both Dropout and L1 Regularization (CNN_dL1) and one deep CNN model; Xception, individually. We aim to provide an efficient and reliable means of categorizing MCM signals, with practical applications in signal processing and communication systems. Extensive simulations demonstrate the effectiveness of our approach, achieving remarkable accuracies. Notably, the Xception model exhibits the highest accuracy among the four models considered. Specifically, we attain an accuracy of 98% at 10 dB SNR using the Xception model. These results underscore the efficacy of our proposed methodology and highlight the potential for its deployment in real-world scenarios.

Efficient construction and convergence analysis of sparse convolutional neural networks

In this paper, a new variant of convolutional neural network (CNN) based on L1 regularization is proposed. The main consideration is to generate a sparse weight matrix, i.e., to generate a sparse neural network model. To some extent, L1 regularization prevents the occurrence of overfitting and effectively improves the learning efficiency of the network. Unfortunately, the L1 regularization term is non-smooth, which leads to the occurrence of oscillations in experiments and poses a great challenge to the theoretical analysis. So, we overcome this drawback by using polishing techniques and rigorously prove the monotonicity of the error function and the strong and weak convergence theorems of the algorithm. Finally, numerical simulations on several data sets support our theoretical results and the superiority of the proposed algorithm.

Remote Sensing and Machine Learning for Safer Railways: A Review

Regular railway inspections are crucial for maintaining their safety and efficiency. However, traditional inspection methods are complex and expensive. Consequently, there has been a significant shift toward combining remote sensing (RS) and machine learning (ML) techniques to enhance the efficiency and accuracy of railway defect monitoring while reducing costs. The advantages of RS-ML techniques include their ability to automate and refine inspection processes and address challenges such as image quality and methodological limitations. However, the integration of RS and ML in railway monitoring is an emerging field, with diverse methodologies and outcomes that the research has not yet synthesized. To fill this gap, this study conducted a systematic literature review (SLR) to consolidate the existing research on RS-ML applications in railway inspection. The SLR meticulously compiled and analyzed relevant studies, evaluating the evolution of research trends, methodological approaches, and the geographic distribution of contributions. The findings showed a notable increase in relevant research activity over the last five years, highlighting the growing interest in this realm. The key methodological patterns emphasize the predominance of approaches based on convolutional neural networks, a variant of artificial neural networks, in achieving high levels of precision. These findings serve as a foundational resource for academics, researchers, and practitioners in the fields of computer science, engineering, and transportation to help guide future research directions and foster the development of more efficient, accurate, and cost-effective railway inspection methods.

Advancing blast fragmentation simulation of RC slabs: A graph neural network approach

Accurate prediction of blast-induced fragmentation in reinforced concrete (RC) structures is pivotal for structural debris hazard assessment in an explosion event. This assessment is essential for designing effective measures for protection of people and structure in the vicinity. While conventional computational techniques are proficient, they are computationally very demanding and may not be readily adaptable in practical applications. This study proposes a novel approach, the Fragment Graph Network (FGN), a specialized variant of graph neural networks (GNNs), to simulate complex interactions between macro-level particles of RC structures subjected to close-in blast loads. The RC structure is discretised into particles, with the model predicting structural responses and fragmentation based on states of these particles. Utilizing an autoregressive prediction mechanism, our model can accurately simulate spatiotemporal responses of RC slabs against blast loads. The efficacy of the model was tested on a dataset consisting of 149 simulated cases, encompassing around 180,000 states each, covering a wide range of blasting scenarios and structural design parameters including TNT charge weight, standoff distance, slab dimension, concrete strength, and reinforcement ratio and layout. Notably, the proposed model delivered accurate predictions, achieving relative errors under 10% and R2 values exceeding 0.93 for both standard and extrapolated test scenarios as compared to the SPH simulations. Remarkably, our model dramatically outpaced traditional numerical approaches, reducing simulation times from hours to minutes on CPUs and seconds on GPUs. This pioneering effort paves the way for efficient and accurate simulations of RC blast fragmentations, offering a rapid alternative to standard numerical approaches.

Variational Neural Network Optimized with Lyrebird Optimization Algorithm for Analysis and Empirical Evidence of Using Data Mining in Financial Risk Prevention Research

In the contemporary landscape of global finance, the effective prevention and mitigation of financial risks have become pivotal for the stability, resilience, and sustainable growth of financial institutions and markets. As financial instruments and markets continue to evolve in complexity, the need for a thorough analysis of risk prevention strategies has become increasingly apparent. In this manuscript, Variational Neural Network (VNN) optimized with lyrebird optimization algorithm (VNN-LOA) is proposed. Initially data is taken from machinehack-financial risk prediction. Afterward the data is fed to Adaptive Variational Bayesian Filter (VBF) based pre-processing process. The pre-processing output is provided to the Variational Neural Network (VNN) to effectively classify financial risk prevention as either involving risk or no risk. The learnable parameters of the VNN are optimized using lyrebird optimization algorithm (LOA). The proposed method is implemented in MATLAB and the efficiency of the proposed method DM-FRP-VNN-COA is estimated with the help of several performances evaluating metrics like, accuracy, precision, recall, f1-score, sensitivity, specificity, computational time, and ROC are analyzed. The proposed DM-FRP-VNN-LOA method attains 38.88%, 35.75%, and 33.16% higher accuracy for risk classification; 34.31%, 38.47% and 37.23% higher accuracy for no risk classification; 22.63%, 33.27% and 21.49% high precision for risk classification; 22.63%, 33.27% and 21.49% high precision for no risk; 35.136%, 39.04% and 38.81% lower computation Time compared with the existing method like data mining using financial risk prevention based back propagation neural network (DM-FRP-BPNN), data mining using financial risk prevention based machine learning (DM-FRP-ML), and data mining using financial risk prevention based convolutional neural network (DM-FRP-CNN) respectively.

SERT-StructNet: Protein secondary structure prediction method based on multi-factor hybrid deep model

Protein secondary structure prediction (PSSP) is a pivotal research endeavour that plays a crucial role in the comprehensive elucidation of protein functions and properties. Current prediction methodologies are focused on deep-learning techniques, particularly focusing on multi-factor features. Diverging from existing approaches, in this study, we placed special emphasis on the effects of amino acid properties and protein secondary structure propensity scores (SSPs) on secondary structure during the meticulous selection of multi-factor features. This differential feature-selection strategy results in a distinctive and effective amalgamation of the sequence and property features. To harness these multi-factor features optimally, we introduced a hybrid deep feature extraction model. The model initially employs mechanisms such as dilated convolution (D-Conv) and a channel attention network (SENet) for local feature extraction and targeted channel enhancement. Subsequently, a combination of recurrent neural network variants (BiGRU and BiLSTM), along with a transformer module, was employed to achieve global bidirectional information consideration and feature enhancement. This approach to multi-factor feature input and multi-level feature processing enabled a comprehensive exploration of intricate associations among amino acid residues in protein sequences, yielding a Q3 accuracy of 84.9% and an Sov score of 85.1%. The overall performance surpasses that of the comparable methods. This study introduces a novel and efficient method for determining the PSSP domain, which is poised to deepen our understanding of the practical applications of protein molecular structures.