Prediction Task Research Articles

Prediction of Disease in Apple Leaf Using Convolution Neural Networks

Abstract: Apple tree diseases can significantly impact fruit quality and yield, making early detection and intervention crucial for orchard management. To avoid the impact on apple production a novel approach is required for the early and accurate prediction of apple leaf diseases through Convolutional Neural Networks (CNNs). Our proposed system contains a comprehensive dataset of high-resolution pictures of apple leaves exhibiting various disease symptoms, including common issues like apple scab, apple rust, and powdery mildew. The dataset was carefully annotated to train and validate the CNN model effectively. The proposed CNN model makes use of its ability to automatically learn applicable features from images, making it highly suitable for the task of disease prediction in apple leaves. This model contributes to the field of precision agriculture by offering a cost-effective and efficient tool for apple disease detection.

Reconstruction of Radio Environment Map Based on Multi-Source Domain Adaptive of Graph Neural Network for Regression

The graph neural network (GNN) has shown outstanding performance in processing unstructured data. However, the downstream task performance of GNN strongly depends on the accuracy of data graph structural features and, as a type of deep learning (DL) model, the size of the training dataset is equally crucial to its performance. This paper is based on graph neural networks to predict and complete the target radio environment map (REM) through multiple complete REMs and sparse spectrum monitoring data in the target domain. Due to the complexity of radio wave propagation in space, it is difficult to accurately and explicitly construct the spatial graph structure of the spectral data. In response to the two above issues, we propose a multi-source domain adaptive of GNN for regression (GNN-MDAR) model, which includes two key modules: (1) graph structure alignment modules are used to capture and learn graph structure information shared by cross-domain radio propagation and extract reliable graph structure information for downstream reference signal receiving power (RSRP) prediction task; and (2) a spatial distribution matching module is used to reduce the feature distribution mismatch across spatial grids and improve the model’s ability to remain domain invariant. Based on the measured REMs dataset, the comparative results of simulation experiments show that the GNN-MDAR outperforms the other four benchmark methods in accuracy when there is less RSRP label data in the target domain.

Quantitative Study of Random Walk Parameters in Node2vec Model

Abstract In this paper, the Monte Carlo simulation method is used to investigate a generalized random walk model based on node2vec which is a popular algorithm in network embedding and has been widely applied in various domains such as link prediction, node classification, recommendation systems, etc. The aim is to quantitatively study the impact of the random walk parameters(including the number of walks per initial node r, the length of each walk l, the return parameter α, the common neighbor parameter β, and the outgoing parameter γ) on the embedding results. Specifically, the cross entropy is utilized as an observation to compare the difference between the frequency of nodes after random walks and the normalized degree sequence of nodes. The results show that the clustering coefficient significantly impacts the cross entropy. For networks with high clustering coefficient, the value of β should closely approximate that of γ, whereas for networks with low clustering coefficient, the value of β should be significantly smaller than that of γ. The value of α should be less than or equal to the minimum values between β and γ. Finally, the embedding effects of different random walk parameters are tested using node classification and link prediction tasks in real-world networks, and the results indicate that cross entropy can provide guidance for obtaining high-quality node embedding.

Traffic prediction via clustering and deep transfer learning with limited data

AbstractThis paper proposes a method based on the clustering algorithm, deep learning, and transfer learning (TL) for short‐term traffic prediction with limited data. To address the challenges posed by limited data and the complex and diverse traffic patterns observed in traffic networks, we propose a profile model based on few‐shot learning to extract each detector's unique profiles. These profiles are then used to cluster detectors with similar patterns into distinct clusters, facilitating effective learning with limited data. A Convolutional Neural Network ‐ Long Short‐Term Memory (CNN‐LSTM)‐based predictive model is proposed to learn and predict traffic volumes for each detector within a cluster. The proposed method demonstrates resilience to detector failures and has been validated using the Performance Measurement System dataset. In scenarios with less than 2 months of training data and 10% failed detectors, the mean absolute error (MAE), root mean square error (RMSE), and mean absolute percentage error (MAPE) for station‐level traffic volume prediction increase from 12.7 vehs/5 min, 20.9 vehs/5 min, and 10.5% to 13.9 vehs/5 min, 24.2 vehs/5 min, and 11.7%, respectively. For lane‐level traffic volume prediction, the average MAE, RMSE, and MAPE increase from 4.7 vehs/5 min, 7.7 vehs/5 min, and 15% to 5.6 vehs/5 min, 9.6 vehs/5 min, and 16.8%. Furthermore, the proposed method extends its applicability to traffic speed and occupancy prediction tasks. TL is integrated to improve speed/occupancy prediction accuracy by leveraging knowledge obtained from traffic volume, considering the correlation between traffic flow, speed, and occupancy. When less than 1 month of traffic speed/occupancy data is available for learning, the proposed method achieves an MAE, RMSE, and MAPE of 0.7 km/h, 1.3 km/h, and 1.3% for station‐level traffic speed prediction and 0.5%, 1.1%, and 11% for station‐level traffic occupancy.

Semi-Supervised Predictive Clustering Trees for (Hierarchical) Multi-Label Classification

Semi-supervised learning (SSL) is a common approach to learning predictive models using not only labeled, but also unlabeled examples. While SSL for the simple tasks of classification and regression has received much attention from the research community, this is not the case for complex prediction tasks with structurally dependent variables, such as multi-label classification and hierarchical multi-label classification. These tasks may require additional information, possibly coming from the underlying distribution in the descriptive space provided by unlabeled examples, to better face the challenging task of simultaneously predicting multiple class labels. In this paper, we investigate this aspect and propose a (hierarchical) multi-label classification method based on semi-supervised learning of predictive clustering trees, which we also extend towards ensemble learning. Extensive experimental evaluation conducted on 24 datasets shows significant advantages of the proposed method and its extension with respect to their supervised counterparts. Moreover, the method preserves interpretability of classical tree-based models.

Cross-tokamak Disruption Prediction based on domain adaptation

Abstract The high acquisition cost and the significant demand for disruptive discharges for data-driven disruption prediction models in future tokamaks pose an inherent contradiction in disruption prediction research. In this paper, we demonstrated a novel approach to predict disruption in a future tokamak using only a few discharges. The approach aims to predict disruption by finding a feature space that is universal to all tokamak. The first step is to use the existing understanding of physics to extract physics-guided features from the diagnostic signals of each tokamak, called physics-guided feature extraction (PGFE). The second step is to align a few data from the future tokamak (target domain) and a large amount of data from existing tokamak (source domain) based on a domain adaptation algorithm called CORrelation ALignment (CORAL). It is the first attempt at applying domain adaptation in the task of cross-tokamak disruption prediction. PGFE has been successfully applied in J-TEXT to predict disruption with excellent performance. PGFE can also reduce the data volume requirements due to extracting the less device-specific features, thereby establishing a solid foundation for cross-tokamak disruption prediction. We have further improved CORAL (supervised CORAL, S-CORAL) to enhance its appropriateness in feature alignment for the disruption prediction task. To simulate the existing and future tokamak case, we selected J-TEXT as the existing tokamak and EAST as the future tokamak, which has a large gap in the ranges of plasma parameters. The utilization of the S-CORAL improves the disruption prediction performance on future tokamak. Through interpretable analysis, we discovered that the learned knowledge of the disruption prediction model through this approach exhibits more similarities to the model trained on large data volumes of future tokamak. This approach provides a light, interpretable and few data-required way by aligning features to predict disruption using small data volume from the future tokamak.

Learning with privileged knowledge of multiple kernels via joint prediction for CT Kernel conversion.

Most existing models for CT kernel conversion take images reconstructed with a single predetermined source kernel as input and convert them to images that are reconstructed with a target kernel. However, these models can achieve even better performance if they leverage complementary information obtained from images reconstructed with multiple different kernels. In many clinical practice scenarios, only images with one kernel can beacquired. We propose a privileged knowledge learning framework that learns privileged knowledge of other source kernels available only in the training data (called privileged information) to guide the conversion from a specific single source kernel to the target kernel, via a joint prediction (JP)task. We construct an ensemble of kernel-specific (KS) tasks where a KS network (KSNet) takes images reconstructed with a specific source kernel as input and converts them to images reconstructed with the target kernel. Then, a JP task is designed to provide extra regularization, which helps each KSNet learn more informative feature representations for kernel conversion, such as detail and structure representations. Meanwhile, we use a cross-shaped window-based attention mechanism in the JP task to highlight the most relevant features to strengthen privileged knowledge learning, thereby alleviating the problems of redundant noise unrelated to images reconstructed with target kernel and inconsistent features that arise from images reconstructed with different kernels. All KSNets can be trained collaboratively by using a JP task to improve the performance of each individual KSNet. We extensively evaluate our method on a clinical dataset with scanners from three manufacturers, that is, Siemens, GE and Philips. The experimental results demonstrate that our privileged knowledge learning framework is effective in improving CT kernel conversion. Through both quantitative and qualitative research, our privileged knowledge learning framework improves the kernel conversion results, thereby contributing to the improvement of diagnostic accuracy and the advancement of comparative research in quantitativemeasurements.

Adaptive Residual Convolutional Neural Network for Compressive Strength Prediction of Energetic Materials Using SEM Images

Abstract The microstructure of materials has a great influence on the macroscopic mechanical properties of materials, and the relationship between them is of great significance to the design of materials. With the development of artificial intelligence and deep learning, many researchers have used convolutional neural networks(CNN) to correlate microscopic images with material properties, and achieved good results. However, for different types of material performance prediction tasks and datasets of different sizes, researchers need to design a special CNN network structure, and the adjustment steps of network structure parameters are cumbersome and time consuming. In this paper, an adaptive residual convolutional neural network with variable depth and width is proposed. We take the prediction error of the test dataset of the specific material performance prediction task as the evaluation index, and adaptively adjusts the network structure based on simulated annealing algorithm to find the best network structure. Finally, it is verified on the dataset composed of SEM images and compressive strength of polymer explosive materials. MAPE of our model is 1.3% and the goodness of fit is 0.943. Further, we use Grad-CAM to explore the correlation between material microstructure and material properties.

Machine Learning in Coastal Engineering: Applications, Challenges, and Perspectives

The integration of machine learning (ML) techniques in coastal engineering marks a paradigm shift in how coastal processes are modeled and understood. While traditional empirical and numerical models have been stalwarts in simulating coastal phenomena, the burgeoning complexity and computational demands have paved the way for data-driven approaches to take center stage. This review underscores the increasing preference for ML methods in coastal engineering, particularly in predictive tasks like wave pattern prediction, water level fluctuation, and morphology change. Although the scope of this review is not exhaustive, it aims to spotlight recent advancements and the capacity of ML techniques to harness vast datasets for more efficient and cost-effective simulations of coastal dynamics. However, challenges persist, including issues related to data availability and quality, algorithm selection, and model generalization. This entails addressing fundamental questions about data quantity and quality, determining optimal methodologies for specific problems, and refining techniques for model training and validation. The reviewed literature paints a promising picture of a future where ML not only complements but significantly enhances our ability to predict and manage the intricate dynamics of coastal environments.

FETILDA: An Evaluation Framework for Effective Representations of Long Financial Documents

In the financial sphere, there is a wealth of accumulated unstructured financial data, such as the textual disclosure documents that companies submit on a regular basis to regulatory agencies, such as the Securities and Exchange Commission (SEC). These documents are typically very long and tend to contain valuable soft information about a company’s performance that is not present in quantitative predictors. It is therefore of great interest to learn predictive models from these long textual documents, especially for forecasting numerical key performance indicators (KPIs). In recent years, there has been a great progress in natural language processing via pre-trained language models (LMs) learned from large corpora of textual data. This prompts the important question of whether they can be used effectively to produce representations for long documents, as well as how we can evaluate the effectiveness of representations produced by various LMs. Our work focuses on answering this critical question, namely the evaluation of the efficacy of various LMs in extracting useful soft information from long textual documents for prediction tasks. In this paper, we propose and implement a deep learning evaluation framework that utilizes a sequential chunking approach combined with an attention mechanism. We perform an extensive set of experiments on a collection of 10-K reports submitted annually by US banks, and another dataset of reports submitted by US companies, in order to investigate thoroughly the performance of different types of language models. Overall, our framework using LMs outperforms strong baseline methods for textual modeling as well as for numerical regression. Our work provides better insights into how utilizing pre-trained domain-specific and fine-tuned long-input LMs for representing long documents can improve the quality of representation of textual data, and therefore, help in improving predictive analyses.

Molecular hypergraph neural networks.

Graph neural networks (GNNs) have demonstrated promising performance across various chemistry-related tasks. However, conventional graphs only model the pairwise connectivity in molecules, failing to adequately represent higher order connections, such as multi-center bonds and conjugated structures. To tackle this challenge, we introduce molecular hypergraphs and propose Molecular Hypergraph Neural Networks (MHNNs) to predict the optoelectronic properties of organic semiconductors, where hyperedges represent conjugated structures. A general algorithm is designed for irregular high-order connections, which can efficiently operate on molecular hypergraphs with hyperedges of various orders. The results show that MHNN outperforms all baseline models on most tasks of organic photovoltaic, OCELOT chromophore v1, and PCQM4Mv2 datasets. Notably, MHNN achieves this without any 3D geometric information, surpassing the baseline model that utilizes atom positions. Moreover, MHNN achieves better performance than pretrained GNNs under limited training data, underscoring its excellent data efficiency. This work provides a new strategy for more general molecular representations and property prediction tasks related to high-order connections.

Explicitly Representing Syntax Improves Sentence-to-Layout Prediction of Unexpected Situations

Abstract Recognizing visual entities in a natural language sentence and arranging them in a 2D spatial layout require a compositional understanding of language and space. This task of layout prediction is valuable in text-to-image synthesis as it allows localized and controlled in-painting of the image. In this comparative study it is shown that we can predict layouts from language representations that implicitly or explicitly encode sentence syntax, if the sentences mention similar entity-relationships to the ones seen during training. To test compositional understanding, we collect a test set of grammatically correct sentences and layouts describing compositions of entities and relations that unlikely have been seen during training. Performance on this test set substantially drops, showing that current models rely on correlations in the training data and have difficulties in understanding the structure of the input sentences. We propose a novel structural loss function that better enforces the syntactic structure of the input sentence and show large performance gains in the task of 2D spatial layout prediction conditioned on text. The loss has the potential to be used in other generation tasks where a tree-like structure underlies the conditioning modality. Code, trained models, and the USCOCO evaluation set are available via Github.1

Robust Long-Term Vehicle Trajectory Prediction Using Link Projection and a Situation-Aware Transformer

The trajectory prediction of a vehicle emerges as a pivotal component in Intelligent Transportation Systems. On urban roads where external factors such as intersections and traffic control devices significantly affect driving patterns along with the driver’s intrinsic habits, the prediction task becomes much more challenging. Furthermore, long-term forecasting of trajectories accumulates prediction errors, leading to substantially inaccurate predictions that may deviate from the actual road. As a solution to these challenges, we propose a long-term vehicle trajectory prediction method that is robust to error accumulation and prevents off-road predictions. In this study, the Transformer model is utilized to analyze and forecast vehicle trajectories. In addition, we propose an extra encoding network to precisely capture the effect of the external factors on the driving pattern by producing an abstract representation of the situation nearby the driver. To avoid off-road predictions, we propose a post-processing method, called link projection, which projects predictions onto the road geometry. Moreover, to overcome the limitations of Euclidean distance-based evaluation metrics in evaluating the accuracy of the entire trajectory, we propose a new metric called area-between-curves (ABC). It measures the similarity between two trajectories, and thus the accordance between the two can be effectively evaluated. Extensive evaluations are conducted using real-world datasets against widely-used methods to demonstrate the effectiveness of the proposed approach. The results show that the proposed approach outperforms the conventional deep learning models by up to 65.74% (RMSE), 60.13% (MAE) and 91.45% (ABC).

Deep Learning Algorithms for IoT Based Crop Yield Optimization

Precision agriculture, with its objectives of optimizing crop yields, decreasing resource waste, and enhancing overall farm management, has emerged as a revolutionary technology in modern agricultural practices. The advent of deep learning techniques and the Internet of Things (IoT) has brought about a paradigm shift in monitoring, decision-making, and predictive analysis within the agriculture industry. This review paper investigates the relationship between deep learning, the (IoT), and agriculture, with an emphasis on how these three domains might work together to optimize crop yields through intelligent decision-making. The integration of deep learning techniques with  (IoT) technology for precision agriculture is thoroughly analyzed in this study, covering recent developments, obstacles, and possible solutions. The paper investigates the role of deep learning algorithms in analyzing the vast amounts of data generated by IoT devices in agriculture. It scrutinizes various deep learning models such as convolutional neural networks (CNNs), recurrent neural networks (RNNs), and their variants applied for crop disease detection, yield prediction, weed identification, and other crucial tasks. Furthermore, this review critically examines the integration of IoT-generated data with deep learning models, highlighting the synergistic benefits in enhancing agricultural decision-making, resource allocation, and predictive analytics. This review underscores the pivotal role of IoT and deep learning techniques in revolutionizing precision agriculture. It emphasizes the need for interdisciplinary collaboration among agronomists, data scientists, and engineers to harness the full potential of these technologies for sustainable and efficient farming practices.

Early prediction of long hospital stay for Intensive Care units readmission patients using medication information

ObjectivePredicting Intensive Care Unit (ICU) Length of Stay (LOS) accurately can improve patient wellness, hospital operations, and the health system's financial status. This study focuses on predicting the prolonged ICU LOS (≥3 days) of the 2nd admission, utilizing short historical data (1st admission only) for early-stage prediction, as well as incorporating medication information. Materials and methodsWe selected 18,572 ICU patients’ records from the MIMIC-IV database for this study. We applied five machine learning classifiers: Logistic regression (LR), Random Forest (RF), Support Vector Machine (SVM), AdaBoost (AB) and XGBoost (XGB). We computed both the sum dose and the average dose for the medication and included them in our model. ResultsThe performance of the RF model demonstrates the highest level of accuracy compared to other models, as indicated by an Area Under the Receiver Operating Characteristic Curve (AUC) of 0.716 and an Expected Calibration Error (ECE) of 0.023. DiscussionThe calibration improved all five classifiers (LR, RF, SVC, AB, XGB) in terms of ECE. The most important two features for RF are the length of 1st admission and the patient's age when they visited the hospital. The most important medication features are Phytonadione and Metoprolol Succinate XL. Also, both the sum and the average dose for the medication features contributed to the prediction task. ConclusionOur model showed the capability to predict the prolonged ICU LOS of the 2nd admission by utilizing the demographic, diagnosis, and medication information from the 1st admission. This method can potentially support the prevention of patient complications and enhance resource allocation in hospitals.

The rationality of explanation or human capacity? Understanding the impact of explainable artificial intelligence on human-AI trust and decision performance

Artificial intelligence models can process massive amounts of data and surpass human experts in predictions. However, the lack of trust in algorithms sealed in the “black box” is one of the most challenging barriers to taking advantage of AI in human decision-making. Improving algorithm transparency by presenting explanations is one of the most common approaches to curing this. Explainable artificial intelligence (XAI) has been a recent research focus, but most concentrate on explainable algorithm development rather than human factors. Thus, the objective of this study is twofold: (1) to explore whether or not XAI can improve human performance and trust in AI in the competitive tasks of sales prediction, and (2) to reveal the different impact routines XAI on individuals with different task-related capacities. Based on a quasi-experimental study, our results indicate that XAI can improve human decision accuracy in the scenario of sales prediction in cross-border e-commerce. XAI cannot improve self-report trust to AI but can improve behavioral trust. We also found the placebo effect of explanation for relatively low task-related capacity.

A transportation Revitalization index prediction model based on Spatial-Temporal attention mechanism

The global outbreak of COVID-19 has had a substantial impact on various sectors worldwide, including the economy, healthcare, entertainment, policy formulation, and international relations, with the transportation industry being particularly hard-hit. To curb the widespread transmission of the virus, many regions globally have implemented policies and measures to restrict transportation. These actions not only directly affect the transportation industry but also further impose a severe impact on the economy and societal development of various areas. In this context, the Transportation Revitalization Index (TRI) becomes particularly important. It can evaluate the degree of recovery of city traffic conditions after the pandemic, and accurate prediction of TRI can help governments and decision-makers respond more precisely to the challenges that the pandemic brings to the transportation industry. However, existing research primarily focuses on the direct correlation between TRI change data and COVID-19 pandemic data, without fully considering the dynamic spatial correlation features and time dependency features that affect the nonlinear changes of TRI. In light of the above situation, this study proposes a Deep Spatial-Temporal prediction model based on the Attention Mechanism (DeepST-AM). The DeepST-AM deeply integrates historical TRI data with multivariate pandemic information and uses a spatial–temporal attention mechanism to capture the deep and complex spatial–temporal information of urban data. To more accurately capture the long-term complex features of TRI data, this paper designs a Gaussian temporal convolution model dedicated to TRI data. To validate the effectiveness of DeepST-AM, researchers used real data from 29 core cities in China as samples and compared the performance of DeepST-AM with existing multiple methods on TRI prediction tasks. The experimental results showed that compared to other methods, the DeepST-AM proposed in this paper has a significant advantage in the long-term prediction tasks of TRI in terms of performance evaluation, indicator prediction, etc. In summary, this research provides a more accurate and comprehensive prediction model for the traffic recovery status after the pandemic, hoping to provide strong support for future decisions.

Boosting Instagram Visibility with Machine Learning

Boosting Instagram Visibility project aims to enhance the visibility of Instagram posts by leveraging machine learning techniques. We follow a systematic process to analyze data gathered from Instagram posts, including details like post content, engagement metrics, and usage of hashtags. Firstly, we collect relevant data from Instagram through various means, ensuring we capture essential information such as post text, engagement metrics, and user demographics. This serves as the foundation for our analysis. Next, we preprocess the collected data to ensure its quality and consistency. We clean the data, handle missing values, and preprocess text to make it suitable for analysis. We then select or extract meaningful features from the data that are crucial for predicting post visibility. These features could include post content, engagement metrics, timing, and demographic information. With the features identified, we choose appropriate machine learning algorithms to analyze the data. These algorithms are selected based on the nature of the prediction task and the characteristics of the dataset. We train the selected machine learning model using the preprocessed data, splitting it into training and validation sets to evaluate the model's performance. After training, we evaluate the model's performance using metrics such as mean absolute error (MAE) or accuracy. This helps us understand how well the model predicts post reach based on the provided features. Using the trained model, we make predictions on new or unseen data, providing insights into the potential visibility of Instagram posts. Through interpretation and analysis of the model predictions, we gain insights into the factors that contribute most to post visibility, helping users optimize their Instagram strategies. Finally, we iterate on the model and process based on feedback and insights gained, continuously refining our approach to improve prediction accuracy. Key Words: Engagement metrics, Hashtags, Mean absolute error, Post visibility, Optimization strategies

Near-earth asteroids classification using NGBoost classifier

Near-Earth asteroids (NEAs) are celestial bodies that orbit within close to Earth, offering valuable insights into the early solar system's formation and posing potential hazards due to impact events. This work presents a comprehensive overview of NEAs, encompassing their historical significance, characteristics, impact hazards, and prospects. The study outlines the NASA Asteroids Classification Dataset and discusses its importance for research on asteroid classification and risk assessment. Furthermore, the methodology section delineates the utilization of the NGBoost classifier for predictive modeling tasks, detailing data collection, preprocessing, model training, evaluation, and result interpretation. Results from the NGBoost classifier demonstrate high accuracy and performance metrics in classifying asteroids, underscoring its efficacy in advancing asteroid classification efforts and informing planetary defense strategies. NEAs pose a potential threat to our planet, and their classification is essential for understanding their properties and predicting their trajectories accurately. In this research, we explore the application of NGBoost, a powerful gradient-boosting framework, for classifying NEAs based on their orbital and physical characteristics. We present a dataset comprising features extracted from known NEAs and non-NEAs and demonstrate the efficacy of NGBoost in accurately distinguishing between these classes. Our results indicate promising performance metrics with 99.22% accuracy, suggesting that NGBoost holds potential as a valuable tool in asteroid classification.

Predicting popularity trend in social media networks with multi-layer temporal graph neural networks

AbstractPredicting what becomes popular on social media is crucial because it helps us understand future topics and public interests based on massive social data. Previous studies mainly focused on picking specific features and checking past statistic numbers, ignoring the hidden impact of messages passing along the complex relationships among different entities. People talk and connect with others on social media; thus, it is essential to consider how information spreads when studying social media networks. This work proposes a multi-layer temporal graph neural network (GNN) framework for predicting what will be popular on social media networks. This framework takes into account the way information spreads among different entities. The proposed method involves multi-layer relations and temporal information within a sequence of social media network snapshots. It learns the temporal representations of target entities in each snapshot and predicts how the popularity of a particular entity will change in future snapshots. The proposed method is evaluated with real-world data across four popularity trend prediction tasks. The experimental results prove that the proposed method performs better than various baselines, including traditional machine learning regression approaches, prior methods for popularity trend prediction, and other GNN models.