Size Of Dataset Research Articles

Processing Next-Generation Mass Spectrometry Imaging Data: Principal Component Analysis at Scale.

Mass spectrometry imaging (MSI) is constantly improving in spatial resolving power, throughput and mass resolution. Although beneficial, these improvements increase data set size and content. The larger data requires correspondingly fast computer-based analyses. However, these analyses often do not scale well with increased data size. Principal component analysis (PCA) is an important analytical tool commonly used with MSI data; however, most PCA algorithms load and process the entire data set within random access memory (RAM) which is most often insufficient for large data sets. PCA algorithms that use less RAM than the data set exist but are usually much slower or sacrifice precision and are rarely used for MSI data processing. Incremental PCA (IPCA) is an alternative algorithm that avoids large RAM allocations while also preserving speed and analytical precision. Here, we demonstrate and benchmark the use of differing implementations of IPCA, PCA, and commercial software on large and often complex MSI data sets. We show that using an already-published Python-based IPCA algorithm, IPCA can be successfully applied to MSI data sets too large to fit with RAM. Furthermore, our benchmarks demonstrate that, contrary to expectations, IPCA is faster than all other tested PCA implementations on all large data sets that can be directly compared.

Cephalometric landmark annotation using transfer learning: Detectron2 and YOLOv8 baselines on a diverse cephalometric image dataset

BackgroundCephalometric landmark annotation is a key challenge in radiographic analysis, requiring automation due to its time-consuming process and inherent subjectivity. This study investigates the application of advanced transfer learning techniques to enhance the accuracy of anatomical landmarks in cephalometric images, which is a vital aspect of orthodontic diagnosis and treatment planning. MethodsWe assess the suitability of transfer learning methods by employing state-of-the-art pose estimation models. The first framework is Detectron2, with two baselines featuring different ResNet backbone architectures: rcnn_R_50_FPN_3x and rcnn_R_101_FPN_3x. The second framework is YOLOv8, with three variants reflecting different network sizes: YOLOv8s-pose, YOLOv8m-pose, and YOLOv8l-pose. These pose estimation models are adopted for the landmark annotation task. The models are trained and evaluated on the DiverseCEPH19 dataset, comprising 1692 radiographic images with 19 landmarks, and their performance is analyzed across various images categories within the dataset. Additionally, the study is extended to a benchmark dataset of 400 images to investigate how dataset size impacts the performance of these frameworks. ResultsDespite variations in objectives and evaluation metrics between pose estimation and landmark localization tasks, the results are promising. Detectron2's variant outperforms others with an accuracy of 85.89%, compared to 72.92% achieved by YOLOv8's variant on the DiverseCEPH19 dataset. This superior performance is also observed in the smaller benchmark dataset, where Detectron2 consistently maintains higher accuracy than YOLOv8. ConclusionThe noted enhancements in annotation precision suggest the suitability of Detectron2 for deployment in applications that require high precision while taking into account factors such as model size, inference time, and resource utilization, the evidence favors YOLOv8 baselines.

Trends in real-time artificial intelligence methods in sports: a systematic review

This review focuses on the usage of machine learning methods in sports. It closely follows the PRISMA framework for writing systematic reviews. We introduce the broader field of using sensor data for feedback in sport and cite similar reviews, that focus on other aspects of the field. With its focus on machine learning models that use signals from simple sensors, this review covers a very focused area that has not yet been covered by any other review. As described in problem definition, we use well-defined inclusion criteria, we have reviewed 72 papers. They present existing solutions, that use machine learning to extract useful information from data collected using various sensors in sports. To be included, papers had to use machine learning methods using data collected from sensors during sports, had to focus on sports-related applications and the result of machine learning had to be some information that can be used in real-time. We have found that the field is rapidly developing as 46 of the 72 included papers were from the last four years. Furthermore, we have found that the field is moving from using classical machine learning techniques to using deep learning. We analyze which data is used as input for machine learning, and we find that the most commonly used sensor is the accelerometer, closely followed by the gyroscope. The most common sensor platform is using a single wearable sensor, however, the studies that used deep learning, use multiple wearable sensors most often. Dataset sizes of sports papers are relatively small compared to other fields, but datasets are on average slightly larger in studies that use deep learning than in those that do not. We analyze the most common preprocessing methods and find that low-pass filtering and feature extraction are commonly used. We compare different machine learning models and the results of the studies that have tested multiple models on the same data, where we find that deep learning proved to be better than classical machine learning. Most studies show classification accuracy of over 90%, showing that machine learning is a useful tool for the researched problems. We end the review by researching how far the machine learning methods were implemented. Twenty of the included papers used their machine learning models in applications beyond a research paper and provided some sort of feedback back to athletes or coaches. After completing the review of the field, we propose a solution – a plan for future research. The proposed solution is to use a combination of best practices from the included paper and methods that we found are not yet implemented in the field of sports. We further elaborate, where we see the current state of the field. We conclude the article with short summary of the findings.

Occluded facial expression recognition based on deep learning

Abstract. When employing Convolutional Neural Networks (CNNs) for facial expression recognition, several challenges are often encountered, such as facial occlusions, the limited size of reliable expression datasets, and inadequate precision in recognition outcomes. This paper preprocesses the dataset to enhance its reliability. By leveraging data synthesis and augmentation techniques, it employs a method of randomly generating occlusion blocks to integrate and expand the dataset. Based on the ResNet-18 network, the model is optimized by incorporating an attention mechanism, thereby improving the network's precision and robustness in recognizing facial expressions.

Image classification with deconvolution operation and augmentation

Abstract Several image classification approaches have been evolved over the years utilizing convolutional neural network (CNN). In convolution operation of CNN, the shifting of kernels to overlapping regions of the image learns redundant data as the images are strongly correlated in reality. The redundant data make the neural network training a challenging task. Again, Deep Learning methods evaluated on small dataset yields degraded performance. To deal with these issues, a proposal is made in this paper that uses deconvolution operation to minimize correlations from images and data augmentation technique to increase the size of datasets. Plant Village, Tomato, and Covid-19 datasets were used for evaluating the performance of the proposed method. 70% of the datasets were used for training, 10% for validation, and 20% for testing purposes. The CIFAR10, MNIST, and Mini-ImageNet datasets were also considered for performance evaluation. The proposed method performed better than other existing methods in terms of classification accuracy.

Evaluating Convolutional Neural Network Architecture for Historical Topographic Hardcopy Maps Analysis: A Study on Training and Validation Accuracy Variation

Convolutional Neural Networks (CNN) are widely used for image analysis tasks, including object detection, segmentation, and recognition. Given the advanced capability, this study evaluates the effectiveness and performance of CNN architecture for analysing Historical Topographic Hardcopy Maps (HTHM) by assessing variations in training and validation accuracy. The lack of research specifically dedicated to CNN’s application in analysing topographic hardcopy maps presents an opportunity to explore and address the unique challenges associated with this domain. While existing studies have predominantly focused on satellite imagery, this study aims to uncover valuable insights, patterns, and characteristics inherent to HTHM through customised CNN approaches. This study utilises a standard CNN architecture and tests the model’s performance with different epoch settings (20, 40, and 60) using varying dataset sizes (288, 636, 1144, and 1716 images). The results indicate that the optimal operation point for training and validation accuracy is achieved at epoch 40. Beyond epoch 40, the widening gap between training and validation accuracy suggests overfitting. Hence, adding more epochs does not significantly improve accuracy beyond the optimum phase. The experiment also shows that the CNN model obtains a training accuracy of 98%, validation accuracy of 67%, and F1-score overall performance of 77%. The analysis demonstrates that the CNN model performs reasonably well in classifying instances from the HTHM dataset. These findings contribute to a better understanding of the strengths and limitations of the model, providing valuable insights for future research and refinement of classification approaches in the context of topographic hardcopy map analysis.

SChemNET: a deep learning framework for predicting small molecules targeting microRNA function

MicroRNAs (miRNAs) have been implicated in human disorders, from cancers to infectious diseases. Targeting miRNAs or their target genes with small molecules offers opportunities to modulate dysregulated cellular processes linked to diseases. Yet, predicting small molecules associated with miRNAs remains challenging due to the small size of small molecule-miRNA datasets. Herein, we develop a generalized deep learning framework, sChemNET, for predicting small molecules affecting miRNA bioactivity based on chemical structure and sequence information. sChemNET overcomes the limitation of sparse chemical information by an objective function that allows the neural network to learn chemical space from a large body of chemical structures yet unknown to affect miRNAs. We experimentally validated small molecules predicted to act on miR-451 or its targets and tested their role in erythrocyte maturation during zebrafish embryogenesis. We also tested small molecules targeting the miR-181 network and other miRNAs using in-vitro and in-vivo experiments. We demonstrate that our machine-learning framework can predict bioactive small molecules targeting miRNAs or their targets in humans and other mammalian organisms.

W1-Net: a highly scalable ptychography convolutional neural network

X-ray ptychography is a coherent diffraction imaging technique that allows for the quantitative retrieval of both the amplitude and phase information of a sample in diffraction-limited resolution. However, traditional reconstruction algorithms require a large number of iterations to obtain phase and amplitude images exactly, and the expensive computation precludes real-time imaging. To solve the inverse problem of ptychography data, PtychoNN uses deep convolutional neural networks for real-time imaging. However, its model is relatively simple, and its accuracy is limited by the size of the training dataset, resulting in lower robustness. To address this problem, a series of W-Net neural network models have been proposed which can robustly reconstruct the object phase information from the raw data. Numerical experiments demonstrate that our neural network exhibits better robustness, superior reconstruction capabilities and shorter training time with high-precision ptychography imaging.

Predictive Regression Modeling for Forecasting Graduation Duration in Online Offsite Degree Program

The demand for Information Technology (IT) professionals continues to rise across various sectors, where theyplay vital roles. However, the supply of IT graduates often fails to meet industry needs and this is a huge problem for the SriLankan IT Industry (National IT-BPM Workforce Survey – 2019). In this context, this study presents a predictive regressionmodelling approach to predict graduation duration in the Bachelor of Information Technology (BIT) degree program at theUniversity of Moratuwa, Sri Lanka. It integrates demographic data—student district, birth year, AL results, OL maths grade,gender, employability status, occupation, and AL stream—along with academic performance indicators like diplomacompletions and higher diploma completions. After evaluating the suggested features, the key findings indicate thesignificance of certain features, notably the number of semesters taken to complete the diploma, higher diploma, and thedegree. Additionally, demographic factors such as district, birth year, AL results, OL maths grade, gender, and employabilitystatus were found to be important. The regression analysis was carried out using the Orange data mining tool (Orange DataMining). Various algorithms, including random forest, neural network, linear regression, and k-nearest neighbours (kNN),were used to develop predictive models. By adjusting parameters such as metrics, weights, number of neighbours, numberof iterations, and training dataset size, the models were optimised to better fit the dataset. Training and testing the modelsrevealed consistent error metrics, including MSE, RMSE, MAE, and R^2, validating the accuracy of predictions. Byconsidering the least and reasonable error in each model, the most suitable model to fit the given dataset was selected.The prediction model accurately forecasted graduation duration for subsequent academic batches, demonstrating itseffectiveness in predicting student progress in the program. This research contributes to understanding the factorsinfluencing graduation duration in a distance learning context and provides insights for educational institutions to optimiseprogram planning and student support initiatives. Additionally, it is a good indicator to the companies to gain a betterunderstanding of the availability of future workforce.

Towards Self-Conscious AI Using Deep ImageNet Models: Application for Blood Cell Classification

The exceptional performance of ImageNet competition winners in image classification has led AI researchers to repurpose these models for a whole range of tasks using transfer learning (TL). TL has been hailed for boosting performance, shortening learning time and reducing computational effort. Despite these benefits, issues such as data sparsity and the misrepresentation of classes can diminish these gains, occasionally leading to misleading TL accuracy scores. This research explores the innovative concept of endowing ImageNet models with a self-awareness that enables them to recognize their own accumulated knowledge and experience. Such self-awareness is expected to improve their adaptability in various domains. We conduct a case study using two different datasets, PBC and BCCD, which focus on blood cell classification. The PBC dataset provides high-resolution images with abundant data, while the BCCD dataset is hindered by limited data and inferior image quality. To compensate for these discrepancies, we use data augmentation for BCCD and undersampling for both datasets to achieve balance. Subsequent pre-processing generates datasets of different size and quality, all geared towards blood cell classification. We extend conventional evaluation tools with novel metrics—“accuracy difference” and “loss difference”—to detect overfitting or underfitting and evaluate their utility as potential indicators for learning behavior and promoting the self-confidence of ImageNet models. Our results show that these metrics effectively track learning progress and improve the reliability and overall performance of ImageNet models in new applications. This study highlights the transformative potential of turning ImageNet models into self-aware entities that significantly improve their robustness and efficiency in various AI tasks. This groundbreaking approach opens new perspectives for increasing the effectiveness of transfer learning in real-world AI implementations.

Machine learning and deep learning models for the diagnosis of apical periodontitis: a scoping review.

To assess the existing literature on the use of machine learning (ML) and deep learning (DL) models for diagnosing apical periodontitis (AP) in humans. A scoping review was conducted following the Arksey and O'Malley framework. The PubMed, SCOPUS, and Web of Science databases were searched, focusing on articles using ML/DL approaches for AP diagnosis. No restrictions were applied. Two independent reviewers screened publications and charted data in predefined Excel tables for analysis. Nineteen publications focused on diagnosing AP by identifying periapical radiolucent lesions (PRLs) in dental radiographs were included. The average sensitivity and specificity for reviewed models were 83% and 90%, respectively. Only three studies explored the direct impact of artificial intelligence (AI) assistance on clinicians' diagnostic performance. Both consistently showed improved sensitivity without compromising specificity. Significant variability in dataset sizes, labeling techniques, and algorithm configurations was noticed. Findings affirm AI models' effectiveness and transformative potential in diagnosing AP by improving the accurate detection of periapical radiolucencies using dental radiographs. However, the lack of standardized reporting on crucial aspects of methodology and performance metrics prevents establishing a definitive diagnostic approach using AI. Further studies are needed to expand AI applications in AP diagnosis beyond radiographic analysis. AI can potentially improve diagnostic accuracy in AP diagnosis by enhancing the sensitivity of PRL detection in dental radiographs without compromising specificity.

Publicly Available Dental Image Datasets for Artificial Intelligence.

The development of artificial intelligence (AI) in dentistry requires large and well-annotated datasets. However, the availability of public dental imaging datasets remains unclear. This study aimed to provide a comprehensive overview of all publicly available dental imaging datasets to address this gap and support AI development. This observational study searched all publicly available dataset resources (academic databases, preprints, and AI challenges), focusing on datasets/articles from 2020 to 2023, with PubMed searches extending back to 2011. We comprehensively searched for dental AI datasets containing images (intraoral photos, scans, radiographs, etc.) using relevant keywords. We included datasets of >50 images obtained from publicly available sources. We extracted dataset characteristics, patient demographics, country of origin, dataset size, ethical clearance, image details, FAIRness metrics, and metadata completeness. We screened 131,028 records and extracted 16 unique dental imaging datasets. The datasets were obtained from Kaggle (18.8%), GitHub, Google, Mendeley, PubMed, Zenodo (each 12.5%), Grand-Challenge, OSF, and arXiv (each 6.25%). The primary focus was tooth segmentation (62.5%) and labeling (56.2%). Panoramic radiography was the most common imaging modality (58.8%). Of the 13 countries, China contributed the most images (2,413). Of the datasets, 75% contained annotations, whereas the methods used to establish labels were often unclear and inconsistent. Only 31.2% of the datasets reported ethical approval, and 56.25% did not specify a license. Most data were obtained from dental clinics (50%). Intraoral radiographs had the highest findability score in the FAIR assessment, whereas cone-beam computed tomography datasets scored the lowest in all categories. These findings revealed a scarcity of publicly available imaging dental data and inconsistent metadata reporting. To promote the development of robust, equitable, and generalizable AI tools for dental diagnostics, treatment, and research, efforts are needed to address data scarcity, increase diversity, mandate metadata completeness, and ensure FAIRness in AI dental imaging research.

Semantic Segmentation Uncertainty Assessment of Different U-net Architectures for Extracting Building Footprints

Abstract. Automatic extraction of building footprints from aerial and space imageries has been found ever increasing importance in urban planning, disaster management, and environmental monitoring. However, achieving accurate building footprint extraction poses significant challenges due to diverse building characteristics and their similarities to their background elements. While conventional methods in building footprint extraction have mainly relied on image processing techniques, recent advancements in deep learning, particularly semantic segmentation algorithms like U-Net, have shown promise in addressing these challenges through machine learning. This study explores different depths of the U-Net model for building footprint extraction, aiming to identify the optimum architecture while investigating the semantic uncertainty of the building footprint extraction. Utilizing aerial imagery from cities including Berlin, Paris, Chicago, and Zurich, collected from Google Maps and OpenStreetMap (OSM) data, five U-Net models have been compared with varying depths. In addition, the impact of dataset sizes and learning rates on model performance has been investigated. Results confirmed that the U-Net-32-1024 model achieves the highest intersection over union (IoU), Accuracy, and F1-score. Moreover, increasing the training dataset size leads to significant improvements in model performance with IoU, Accuracy and F1-score reaching their values of 73.73%, 88.65% and 88.53%. However, challenges remain in accurately delineating buildings in dense urban areas. Nonetheless, our findings demonstrated the effectiveness of U-Net models in building footprint extraction.

Enhancing Cardiovascular Risk Prediction: Development of an Advanced Xgboost Model with Hospital-Level Random Effects.

Ensemble tree-based models such as Xgboost are highly prognostic in cardiovascular medicine, as measured by the Clinical Effectiveness Metric (CEM). However, their ability to handle correlated data, such as hospital-level effects, is limited. The aim of this work is to develop a binary-outcome mixed-effects Xgboost (BME) model that integrates random effects at the hospital level. To ascertain how well the model handles correlated data in cardiovascular outcomes, we aim to assess its performance and compare it to fixed-effects Xgboost and traditional logistic regression models. A total of 227,087 patients over 17 years of age, undergoing cardiac surgery from 42 UK hospitals between 1 January 2012 and 31 March 2019, were included. The dataset was split into two cohorts: training/validation (n = 157,196; 2012-2016) and holdout (n = 69,891; 2017-2019). The outcome variable was 30-day mortality with hospitals considered as the clustering variable. The logistic regression, mixed-effects logistic regression, Xgboost and binary-outcome mixed-effects Xgboost (BME) were fitted to both standardized and unstandardized datasets across a range of sample sizes and the estimated prediction power metrics were compared to identify the best approach. The exploratory study found high variability in hospital-related mortality across datasets, which supported the adoption of the mixed-effects models. Unstandardized Xgboost BME demonstrated marked improvements in prediction power over the Xgboost model at small sample size ranges, but performance differences decreased as dataset sizes increased. Generalized linear models (glms) and generalized linear mixed-effects models (glmers) followed similar results, with the Xgboost models also excelling at greater sample sizes. These findings suggest that integrating mixed effects into machine learning models can enhance their performance on datasets where the sample size is small.

Designing the Next Generation of Polymers with Machine Learning and Physics-Based Models

Abstract The development of next-generation polymers necessitates optimizing several key properties simultaneously, a task that is expensive and infeasible using traditional trial-and-error experimental approaches. A promising alternative is employing a combination of machine learning and physics-based tools to rapidly screen the polymer design space and provide suggestions of new polymers that meet the critical properties required for industrial applications. In this study, we introduce a comprehensive workflow that utilizes machine learning and molecular modeling approaches to design new polymers with the focus on improving five polymer properties: (1) glass transition temperature, (2) dielectric constant, (3) refractive index, (4) stress optic coefficient, and (5) linear coefficient of thermal expansion. Using a small dataset (<200 unique polymers), we developed quantitative structure-property relationships (QSPR) models to accurately predict the experimental polymer properties for both homo- and co-polymer systems. We tested several ML algorithms and identified the best models for predicting these polymer properties, achieving test set R2 greater than 0.77 across all properties. We then explored new polymers by creating a library of over ~10,000 homopolymers using R-group enumeration tools and applied the trained QSPR models to rapidly predict the five polymer properties. The predictions of QSPR models were used to create a multi-parameter optimization score, which helped downselect the large polymer space to ~10 promising candidates. The properties of these selected polymer candidates were subsequently validated with classical molecular dynamics simulations and density functional theory, revealing a strong correlation with the QSPR model predictions. Finally, one of the top candidates was validated by experiments, which showed good agreement against QSPR and physics-based models. Our workflow underscores the power of combining data-driven and theoretical methods in the polymer design process given a small dataset size, offering a valuable resource for experimentalists looking to leverage computer-aided strategies in materials innovation.

Automated lesion detection in cotton leaf visuals using deep learning

Cotton is one of the major cash crop in the agriculture led economies across the world. Cotton leaf diseases affects its yield globally. Determining cotton lesions on leaves is difficult when the area is big and the size of lesions is varied. Automated cotton lesion detection is quite useful; however, it is challenging due to fewer disease class, limited size datasets, class imbalance problems, and need of comprehensive evaluation metrics. We propose a novel deep learning based method that augments the data using generative adversarial networks (GANs) to reduce the class imbalance issue and an ensemble-based method that combines the feature vector obtained from the three deep learning architectures including VGG16, Inception V3, and ResNet50. The proposed method offers a more precise, efficient and scalable method for automated detection of diseases of cotton crops. We have implemented the proposed method on publicly available dataset with seven disease and one health classes and have achieved highest accuracy of 95% and F-1 score of 98%. The proposed method performs better than existing state of the art methods.

Chemoinformatics for corrosion science: Data-driven modeling of corrosion inhibition by organic molecules.

This paper reviews the application of machine learning to the inhibition of corrosion by organic molecules. The methodologies considered include quantitative structure-property relationships (QSPR) and related data-driven approaches. The characteristic features of their key components are considered as applied to corrosion inhibition, including datasets, response properties, molecular descriptors, machine learning methods, and structure-property models. It is shown that the most important factors determining their choice and application features are: (1) the small or very small size of datasets, (2) the mechanism of corrosion inhibition associated with the adsorption of inhibitor molecules on the metal surface, and (3) multifactorial conditioning and noisiness of response property. On this basis, the application of machine learning to the inhibition of corrosion of materials based on iron, aluminum, and magnesium is considered. The main trends in the development of QSPR and related data-driven modeling of corrosion inhibition are discussed, the shortcomings and common errors are considered, and the prospects for their further development are outlined.

Improving thyroid cancer diagnosis with advanced deep learning techniques

The accurate diagnosis of thyroid carcinoma is a complex and critical challenge in medical practice due to the clinical and pathological diversity of thyroid tumors. This study explores the application of advanced deep learning models, specifically EfficientNet B4 and MobileNetV3, to enhance the classification of thyroid histopathological images. By leveraging transfer learning, these models were fine-tuned on a dataset of 7,272 images encompassing three types of carcinoma: Medullary, Papillary, and Vesicular. The performance of these models was evaluated using metrics such as accuracy, precision, recall, and F1-score. The results indicate that both models are effective, with EfficientNet B4 demonstrating slightly superior accuracy, particularly in more challenging diagnostic scenarios. Despite these promising results, the study acknowledges limitations related to dataset size and variability, which could affect the generalizability of the models. To address these issues, future research should focus on expanding the dataset and incorporating additional diagnostic data, such as genetic and molecular markers, to improve model performance and clinical applicability. This study highlights the potential of deep learning models to significantly enhance the diagnostic accuracy of thyroid carcinomas, paving the way for more reliable and precise clinical decision-making.

Pilot study for non-invasive diabetes detection through classification of photoplethysmography signals using convolutional neural networks

Diabetes is a chronic disorder affecting vascular health, often altering pulse wave characteristics. Traditional pulse wave analysis (PWA) methods face challenges such as variability and complexity of signals. This study aims to overcome these limitations by leveraging deep learning models for more accurate and efficient classification. The methodology used in this study involves four key steps: data collection, data preprocessing, Convolutional Neural Network (CNN) model development, and model evaluation. Primary data were collected using a multipara patient monitor, including finger photoplethysmography (PPG) signals, blood pressure, mean arterial pressure, oxygen saturation, and pulse rate. Single pulse wave cycles from 60 healthy individuals and 60 patients with type 2 diabetes underwent preprocessing. The CNN model was trained using 50 PPG images from each group and achieved a training accuracy of 92%. The prediction capability of the model was evaluated using 20 unseen images, comprising 10 healthy and 10 diabetes PPG images. It attained a 90% overall test accuracy in distinguishing between PPG images of individuals with diabetes and those who are healthy. These findings suggest that CNN-based analysis of PPG signals provides a precise, non-invasive tool for diabetes screening. To further enhance accuracy, future studies should focus on increasing the dataset size and performing hyperparameter tuning to optimize the CNN model.

Forecasting the equity premium: can machine learning beat the historical average?

We empirically predict the equity premium with the selected machine learning methods in Gu et al. (Empirical asset pricing via machine learning. Rev. Financ. Stud., 2020, 33(5), 2223–2273). We also consider four additional popular machine learning methods (ridge regression, support vector regression, k-nearest neighbors, and extreme gradient boosted trees) and their combination method. Using a dataset of both macroeconomic and technical predictors, we find that despite showcasing strong in-sample forecasting abilities, particularly with tree-based models, the out-of-sample results support Welch and Goyal (A comprehensive look at the empirical performance of equity premium prediction. Rev. Financ. Stud., 2008, 21(4), 1455–1508) that the competing forecasting models generally fail to outperform the historical average benchmark. We attribute this failure to the small dataset size and the low signal-to-noise ratio inherent in equity premium prediction. Our variable importance analysis further identifies three bond interest rate-related variables as the most dominant predictors for the equity premium. The economic value from a market timing perspective highlights that the historical average benchmark strategy generates the highest average return of 25.63% and the best Sharpe ratio of 0.8. Finally, our findings are robust across a variety of settings.