Key Hyperparameters Research Articles

Improving Deep Learning-Based Recommendation Attack Detection Using Harris Hawks Optimization

Recommendation attack attempts to bias the recommendation results of collaborative recommender systems by injecting malicious ratings into the rating database. A lot of methods have been proposed for detecting such attacks. Among these works, the deep learning-based detection methods get rid of the dependence on hand-designed features of recommendation attack besides having excellent detection performance. However, most of them optimize the key hyperparameters by manual analysis which relies too much on domain experts and their experience. To address this issue, in this paper we propose an approach based on the Harris Hawks Optimization (HHO) algorithm to improve the deep learning-based detection methods. Being different from the original detection methods which optimize the key hyperparameters manually, the improved deep learning-based detection methods can optimize the key hyperparameters automatically. We first convert the key hyperparameters of discrete type to continuous type according to the uniform distribution theory to expand the application scope of HHO algorithm. Then, we use the detection stability as an early stop condition to reduce the optimization iterations to improve the HHO algorithm. After that, we use the improved HHO algorithm to automatically optimize the key hyperparameters for the deep learning-based detection methods. Finally, we use the optimized key hyperparameters to train the deep learning-based detection methods to generate classifiers for detecting the recommendation attack. The experiments conducted on two benchmark datasets illustrate that the improved deep learning-based detection methods have effective performance.

An interpretable ensemble-learning-based open source model for evaluating the fire resistance of concrete-filled steel tubular columns

A new simulation model rooted in explainable, pragmatic machine learning theories is proposed which simply and accurately predicts the fire resistance of concrete-filled steel tubular (CFST) columns. The XGBoost, an ensemble-learning capable method, was used to formulate the model and its key hyper-parameters were fine-tuned using the meta-heuristic balancing composite motion optimization (BCMO) algorithm. Model-agnostic approaches were applied to elucidate the underlying physical mechanisms of the developed model. Two explicit fire-resistance design equations derived from generalized linear model and genetic expression programming are also presented and their accuracy is shown to surpass that of conventional fire-resistance prediction methods. Finally, a preliminary reliability analysis based on the proposed model is conducted. A convenient graphic user interface together with all source code files are provided for practical and academic use.

Accurate Compression of Tabulated Chemistry Models with Partition of Unity Networks

ABSTRACT Tabulated chemistry models are widely used to simulate large-scale turbulent fires in applications including energy generation and fire safety. Tabulation via piecewise Cartesian interpolation suffers from the curse-of-dimensionality, leading to a prohibitive exponential growth in parameters and memory usage as more dimensions are considered. Artificial neural networks (ANNs) have attracted attention for constructing surrogates for chemistry models due to their ability to perform high-dimensional approximation. However, due to well-known pathologies regarding the realization of suboptimal local minima during training, in practice they do not converge and provide unreliable accuracy. Partition of unity networks (POUnets) are a recently introduced family of ANNs which preserve notions of convergence while performing high-dimensional approximation, discovering a mesh-free partition of space which may be used to perform optimal polynomial approximation. We assess their performance with respect to accuracy and model complexity in reconstructing unstructured flamelet data representative of nonadiabatic pool fire models. Our results show that POUnets can provide the desirable accuracy of classical spline-based interpolants with the low memory footprint of traditional ANNs while converging faster to significantly lower errors than ANNs. For example, we observe POUnets obtaining target accuracies in two dimensions with 40 to 50 times less memory and roughly double the compression in three dimensions. We also address the practical matter of efficiently training accurate POUnets by studying convergence over key hyperparameters, the impact of partition/basis formulation, and the sensitivity to initialization.

IoT framework for brain tumor detection based on optimized modified ResNet 18 (OMRES)

Brain tumors are a serious health issue that affects many people’s lives. Such a tumor, which is either benign or malignant, can be fatal if malignant cells are not correctly diagnosed. According to the most recent human health care analysis system, the number of brain tumor patients has climbed dramatically and is now the 10th top cause of death. As a result, detecting brain tumors in their early stages can considerably improve the patient’s prospects of complete recovery and therapy. Thanks to improvements in information and communication technology, the Internet of things (IoT) has reached an evolutionary stage in the development of the modern health care environment. This paper provides a detailed examination of brain tumor detection approaches. Moreover, two different scenarios for detecting brain tumors will be proposed. On one hand, the first scenario depends on applying a deep convolutional neural network directly to brain images. On the other hand, the second scenario presents an IoT-based framework that adopts a multiuser detection system by sending the images to the cloud for early detection of brain tumors, which makes the system accessible to anyone and anywhere for accurate brain tumor categorization. The proposed CNN structure can be considered a modified version of the pre-trained ResNet18 CNN. Additionally, two key hyper-parameters are used to fine-tune the OMRES model, firstly different optimizers are tested using different learning rates, batch sizes, and a constant number of epochs, and secondly, the impact of changing dropout rates is made. Finally, comparisons between the OMRES model and traditional pre-trained models are discussed. Based on simulation findings, the RMSProp algorithm with a dropout rate of 0.5 verifies the best outcomes over other algorithms, where the suggested model achieves superior improvement with the highest rated accuracy of 98.67% compared to the conventional CNNs.

Semi-supervised EEG emotion recognition model based on enhanced graph fusion and GCN

Objective. To take full advantage of both labeled data and unlabeled ones, the Graph Convolutional Network (GCN) was introduced in electroencephalography (EEG) based emotion recognition to achieve feature propagation. However, a single feature cannot represent the emotional state entirely and precisely due to the instability of the EEG signal and the complexity of the emotional state. In addition, the noise existing in the graph may affect the performance greatly. To solve these problems, it was necessary to introduce feature/similarity fusion and noise reduction strategies. Approach. A semi-supervised EEG emotion recognition model combining graph fusion, network enhancement, and feature fusion was proposed. Firstly, different features were extracted from EEG and then compacted by Principal Component Analysis (PCA), respectively. Secondly, a Sample-by-sample Similarity Matrix (SSM) was constructed based on each feature, and similarity network fusion (SNF) was adopted to fuse the graphs corresponding to different SSMs to take advantage of their complementarity. Then, Network Enhancement (NE) was performed on the fused graph to reduce the noise in it. Finally, GCN was performed on the concatenated features and the enhanced fused graph to achieve feature propagation. Main results. Experimental results demonstrated that: (a) When 5.30 of SEED and 7.20 of SEED-IV samples were chosen as the labeled samples, respectively, the minimum classification accuracy improvement achieved by the proposed scheme over state-of-the-art schemes were 1.52 on SEED and 13.14 on SEED-IV, respectively. (b) When 8.00 of SEED and 9.60 of SEED-IV samples were chosen as the labeled samples, respectively, the minimum training time reduction achieved by the proposed scheme over state-of-the-art schemes were 46.75 s and 22.55 s, respectively. (c) Graph fusion, network enhancement, and feature fusion all contributed to the performance enhancement. (d) The key hyperparameters that affect the performance were relatively few and easy to set to obtain outstanding performance. Significance. This paper demonstrated that the combination of graph fusion, network enhancement, and feature fusion help to enhance GCN-based EEG emotion recognition.

Effects of hyperparameters on flow field reconstruction around a foil by convolutional neural networks

Unsteady flow field over a foil at Rec = 300 was used to train the convolutional neural network (CNN) model with different hyperparameter settings based on the mode decomposing CNN autoencoder (MD-CNN-AE). The effects of three key hyperparameters, such as the number of convolution kernels, the kernel size and the number of convolution layers of CNN models, on the flow field reconstruction were investigated. The reconstructed error and the energy distribution of the orthogonal basis of the flow field were utilized to evaluate the differences. The decomposed fields generated by the latent vectors in the model and the corresponding mode decomposition performed by the proper orthogonal decomposition (POD) were visualized in the physical space to observe the differences directly. The present results showed that there is no obvious monotonicity between the reconstructed error and the number of kernels; secondly, the kernel size had little effect on the error; however, when considering the number of convolution layers, a conclusion contrary to the general information in the literature is obtained, that the reconstructed error increases with the number of convolution layers. The relationship between the decomposed field and the orthogonal basis was also analyzed.

Stochastic deep collocation method based on neural architecture search and transfer learning for heterogeneous porous media

We present a stochastic deep collocation method (DCM) based on neural architecture search (NAS) and transfer learning for heterogeneous porous media. We first carry out a sensitivity analysis to determine the key hyper-parameters of the network to reduce the search space and subsequently employ hyper-parameter optimization to finally obtain the parameter values. The presented NAS based DCM also saves the weights and biases of the most favorable architectures, which is then used in the fine-tuning process. We also employ transfer learning techniques to drastically reduce the computational cost. The presented DCM is then applied to the stochastic analysis of heterogeneous porous material. Therefore, a three dimensional stochastic flow model is built providing a benchmark to the simulation of groundwater flow in highly heterogeneous aquifers. The performance of the presented NAS based DCM is verified in different dimensions using the method of manufactured solutions. We show that it significantly outperforms finite difference methods in both accuracy and computational cost.

MetaPan: Unsupervised Adaptation With Meta-Learning for Multispectral Pansharpening

Multispectral (MS) pansharpening aims to improve the spatial resolution of MS images (MSI) using the spatial details of panchromatic (PAN) images. Due to the gap of prior knowledge between the simulated data and real-world cases, unsupervised learning-based approaches have grown increasing interest. However, some key hyper-parameters, such as the initial weights of the networks, are set manually, which significantly impacts the fusion performance. To tackle this problem, we propose a novel unsupervised adaptation method with meta-learning for MS pansharpening (MetaPan), in which the meta-learning aims to automatically learn the initial parameters of a three-stream fusion network (TSFNet) for unsupervised adaptation learning (UAL). Specifically, the TSFNet consists of a PAN stream, an MS stream, and a fusion stream, where the fusion stream implicitly leverages domain-specific knowledge of input image pairs while the other two streams explicitly inject spatial details and spectral information into the fusion stream. The MetaPan consists of a pre-training stage, a meta-learning stage, and a UAL stage. At the pre-training stage, the TSFNet is trained with the supervision of simulated ground truth such that it is universal for all image pairs. Then, the process of meta-learning optimizes for an internal representation of network parameters that can adapt to a specific image pair with UAL through only a few steps. Finally, the learned internal representation is fine-tuned to a real-world image pair (a test image pair) with UAL. Experiments on two datasets show that our method performs better than state-of-the-art methods in both quantitative metrics and visual appearance.

Hierarchical Attention-Based Machine Learning Model for Radiation Prediction of WB-BGA Package

Rapid increase in operating frequency of integrated chips and intricacy of electronic packages outpaces the ability of conventional methods in coping with the growing complexity of electromagnetic interference (EMI) issues. To address it, several machine learning (ML) methods––deep neural network (DNN), convolutional neural network, support vector regression, <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">K</i> -nearest neighbor, and linear regression are constructed to acquire the best ML model to accurately and rapidly predict the maximum 3-m radiated electric field of a wire-bond ball grid array package. The key hyperparameters of different ML models are tuned respectively to attain the least prediction error for each model. Among the optimized ML models, the prediction accuracy of the DNN model is the highest. In this article, a hierarchical attention-based DNN model is proposed and discussed in depth to reduce the number of training datasets, and identify the structural parameters with large contributions to radiation prediction. These structural parameters with contributions can guide the packaging design. The DNN model with attention-weight input requires fewer training datasets than the original DNN model. Furthermore, the experimental measurement for EMI radiation of a test package board is implemented, and the far-field results show the effectiveness and feasibility of the DNN model.

Intelligent fault diagnosis of planetary gearbox based on adaptive normalized CNN under complex variable working conditions and data imbalance

In real industrial application, the operating conditions of the planetary gearbox are always variable speed and variable load according to production. Meanwhile, the scarcity of fault data and different working conditions lead to serious data imbalance and distribution differences, which make a great challenge for planetary gearbox fault detection and diagnosis. To address these issues, a novel adaptive normalized convolutional neural network (ANCNN) is developed to accurately and automatically diagnose the different fault locations and severities of planetary gearbox considering the scenarios of complex variable working conditions and data imbalance. First, Teager calculated order (TCO) spectrum is applied to avoid the influences of large speed fluctuations and variable loads, which is an efficient pretreatment approach for the proposed ANCNN. Then, the batch normalization algorithm is adopted to eliminate the feature distribution differences caused by variable operating modes and data imbalance. Finally, in order to automatically adapt to different planetary gearbox diagnosis circumstances, the particle swarm optimization (PSO) strategy is used for optimizing and flexibly deciding the key hyperparameters of the designed model, thereby improving the overall performance of the model. The effectiveness of the presented method is confirmed through experiments on two different planetary gearbox datasets, which are from a generic planetary gearbox for industrial applications and a drivetrain dynamic simulator test rig. In addition, comparison with other mainstream intelligent diagnosis techniques validates the superiority of the presented method. The experimental results demonstrate that the presented method can achieve diagnostic accuracies of better than 99.8%, and also shows excellent stability for the unbalanced data classification.

Towards novel deep neuroevolution models: chaotic levy grasshopper optimization for short-term wind speed forecasting

High accurate wind speed forecasting plays an important role in ensuring the sustainability of wind power utilization. Although deep neural networks (DNNs) have been recently applied to wind time-series datasets, their maximum performance largely leans on their designed architecture. By the current state-of-the-art DNNs, their architectures are mainly configured in manual way, which is a time-consuming task. Thus, it is difficult and frustrating for regular users who do not have comprehensive experience in DNNs to design their optimal architectures to forecast problems of interest. This paper proposes a novel framework to optimize the hyperparameters and architecture of DNNs used for wind speed forecasting. Thus, we introduce a novel enhanced version of the grasshopper optimization algorithm called EGOA to optimize the deep long short-term memory (LSTM) neural network architecture, which optimally evolves four of its key hyperparameters. For designing the enhanced version of GOA, the chaotic theory and levy flight strategies are applied to make an efficient balance between the exploitation and exploration phases of the GOA. Moreover, the mutual information (MI) feature selection algorithm is utilized to select more correlated and effective historical wind speed time series features. The proposed model’s performance is comprehensively evaluated on two datasets gathered from the wind stations located in the United States (US) for two forecasting horizons of the next 30-min and 1-h ahead. The experimental results reveal that the proposed model achieves the best forecasting performance compared to seven prominent classical and state-of-the-art forecasting algorithms.

FedSA: A staleness-aware asynchronous Federated Learning algorithm with non-IID data

This paper presents new asynchronous methods to the Federated Learning (FL), one of the next-generation paradigms for Artificial Intelligence (AI) systems. We consider the two-fold challenges lay ahead. First, non-IID (non-Independent and Identically Distributed) data across devices cause unstable performance. Second, unreliable and slow environments not only slow the convergence but also cause staleness issues. To address these challenges, this study uses a bottom-up approach for analysis and algorithm design. We first reformulate FL by unifying both synchronous and asynchronous updating schemes with an asynchrony-related parameter. We theoretically analyze this new form and find practical strategies for optimization. The key findings include: 1) a two-stage training strategy to accelerate training and reduce communication overhead; 2) strategies of choosing key hyperparameters optimally for these stages to maintain efficiency and robustness. With these theoretical guarantees, we propose FedSA (Federated Staleness-Aware), a novel asynchronous federated learning algorithm. We validate FedSA on different tasks with non-IID/IID and staleness settings. Our results indicate that, given a large proportion of stale devices, the proposed algorithm presents state-of-the-art performance by outperforming existing methods on both non-IID and IID cases.

Dynamic Hand Gesture Recognition Based on Signals From Specialized Data Glove and Deep Learning Algorithms

Gesture recognition as a natural, convenient and recognizable way has been received more and more attention on human–machine interaction (HMI) recently. However, visual-based gesture recognition methods are often restricted by environments and classical wearable device-based strategies are suffered from relatively low accuracy or the complicated structures. In this study, we first design a low-cost and efficient data glove with simple hardware structure to capture finger movement and bending simultaneously. Second, a novel dynamic hand gesture recognition algorithm (DGDL-GR) is proposed to recognize human dynamic sign language, in which a fusion model of convolutional neural network (fCNN) and generic temporal convolutional network (TCN) is fully utilized. The fCNN (fusion of 1-D CNN and 2-D CNN) is proposed to extract time-domain features of finger resistance movement and spatial domain features of finger resistance bending simultaneously. Moreover, due to the superiorities of TCN in sequence modeling task, this work proposes a novel hand gesture recognition method based on the TCN, which includes causal convolution, dilation convolution, and a residual network with appropriate layers. Both long- and short-time dependencies of the hand gesture features are deeply mined and classified in the end. Results of extensive experiments have demonstrated that the proposed DGDL-GR algorithm outperforms many state-of-the-art algorithms on the measure of accuracy, F1 score, precision score, and recall score with the real-world dataset. Moreover, the number of residual blocks and some key hyperparameters of the proposed DGDL-GR algorithm has been studied thoroughly in this work.

Deep Learning in Forest Structural Parameter Estimation Using Airborne LiDAR Data

Accurately estimating and mapping forest structural parameters are essential for monitoring forest resources and understanding ecological processes. The novel deep learning algorithm has the potential to be a promising approach to improve the estimation accuracy while combining with advanced remote sensing technology. Airborne light detection and ranging (LiDAR) has the preferable capability to characterize 3-D canopy structure and estimate forest structural parameters. In this study, we developed a deep learning-based algorithm (Deep-RBN) that combined the fully connected network (FCN) deep learning algorithm with the optimized radial basis neural network (RBN) algorithm for forest structural parameter estimation using airborne LiDAR data. The multiple iterations were used to constantly update the internal weights to achieve the optimized accuracy of model fitting, and the optimized RBN algorithm was developed for the limited training sets. We assessed the efficiency and capability of the Deep-RBN in the estimation of forest structural parameters in a subtropical planted forest of southern China, by comparing the traditional FCN algorithm and multiple linear regression. We found that Deep-RBN had the strongest capability in estimates of forest structural parameters (R2 = 0.67–0.86, rRMSE = 6.95%–20.34%). The sensitivity analysis of the key hyperparameters of Deep-RBN algorithm showed that the learning rate is one of the most important parameters that influence the performance of predictive models, and while its value equal is to 0.001, the predictive models had the highest accuracy (mean DBH: RMSE = 1.01, mean height: RMSE = 1.45, volume: RMSE = 26.49, stem density: RMSE = 121.06). With the increase of training samples added in Deep-RBN model, the predictive models performed better; however, no significant improvements of accuracy were observed while the number of training set is larger than 80. This study demonstrates the benefits of jointly using the Deep-RBN algorithm and airborne LiDAR data to improve the accuracy of forest structural parameter estimation and mapping, which provides a promising methodology for sustainable forest resources monitoring.

Life prediction of lithium-ion batteries based on stacked denoising autoencoders

Accurate life prediction of lithium-ion batteries is important to help assess battery quality in advance, improve long-term battery planning, and subsequently guarantee the safety and reliability of battery operations. In this study, a deep learning-based stacked denoising autoencoder (SDAE) method is proposed to directly predict battery life by extracting various battery features. In general, the SDAE contains autoencoders and uses a deep network architecture to learn the complex nonlinear input-output relationship in a layer-by-layer fashion. Many features enabling the life prediction of lithium-ion batteries are extracted from discharge temperature and voltage curves. As redundancies in these features may result in poor prediction accuracy, a clustering by fast search (CFS) method is adopted to filter and select essential features. The CFS selects effective features by aggregating the types of battery features into clusters. All selected features are then fed into the SDAE to predict battery life cycle. Key hyperparameters are investigated, such as the number of iterations, the learning rate, and the denoising rate of the SDAE network. Experimental results show that the proposed selected-features-based deep learning method can provide more accurate and efficient battery life predictions with less fluctuation than the method without feature selection.

Fault diagnosis with synchrosqueezing transform and optimized deep convolutional neural network: An application in modular multilevel converters

High voltage direct current (HVDC) transmission mode with modular multilevel converters (MMC) topology is the future direction of transmission engineering, and security is their fundamental issue. Submodule fault of MMC in HVDC is the most common problem, nevertheless, traditional time–frequency based diagnosis technology can’t achieve high accuracy. To solve this pain spot, a new diagnosis strategy based on the synchrosqueezing transform (SST) and genetic algorithm optimized deep convolution neural network (GA-DCNN) is proposed in this paper. Firstly, the time–frequency representations (TFRs) of the raw signals which is synthesized by ac current and inner circulating current of the MMC are calculated with SST. Then, DCNN is introduced to learn the underlying features from the TFRs, and its key hyperparameters are optimized with genetic algorithm. Meanwhile, batch normalization, dropout and data augment technologies are explored to prevent DCNN model from overfitting and improve model performance. Compared to traditional SVM and BP-based algorithms, SST-GA-DCNN achieve high diagnosis accuracy. The experimental results show the feasibility and applicability of the proposed fault diagnosis framework.

Multi-Frame Star Image Denoising Algorithm Based on Deep Reinforcement Learning and Mixed Poisson-Gaussian Likelihood.

Mixed Poisson–Gaussian noise exists in the star images and is difficult to be effectively suppressed via maximum likelihood estimation (MLE) method due to its complicated likelihood function. In this article, the MLE method is incorporated with a state-of-the-art machine learning algorithm in order to achieve accurate restoration results. By applying the mixed Poisson–Gaussian likelihood function as the reward function of a reinforcement learning algorithm, an agent is able to form the restored image that achieves the maximum value of the complex likelihood function through the Markov Decision Process (MDP). In order to provide the appropriate parameter settings of the denoising model, the key hyperparameters of the model and their influences on denoising results are tested through simulated experiments. The model is then compared with two existing star image denoising methods so as to verify its performance. The experiment results indicate that this algorithm based on reinforcement learning is able to suppress the mixed Poisson–Gaussian noise in the star image more accurately than the traditional MLE method, as well as the method based on the deep convolutional neural network (DCNN).

Unsupervised dimensionality reduction versus supervised regularization for classification from sparse data

Unsupervised matrix-factorization-based dimensionality reduction (DR) techniques are popularly used for feature engineering with the goal of improving the generalization performance of predictive models, especially with massive, sparse feature sets. Often DR is employed for the same purpose as supervised regularization and other forms of complexity control: exploiting a bias/variance tradeoff to mitigate overfitting. Contradicting this practice, there is consensus among existing expert guidelines that supervised regularization is a superior way to improve predictive performance. However, these guidelines are not always followed for this sort of data, and it is not unusual to find DR used with no comparison to modeling with the full feature set. Further, the existing literature does not take into account that DR and supervised regularization are often used in conjunction. We experimentally compare binary classification performance using DR features versus the original features under numerous conditions: using a total of 97 binary classification tasks, 6 classifiers, 3 DR techniques, and 4 evaluation metrics. Crucially, we also experiment using varied methodologies to tune and evaluate various key hyperparameters. We find a very clear, but nuanced result. Using state-of-the-art hyperparameter-selection methods, applying DR does not add value beyond supervised regularization, and can often diminish performance. However, if regularization is not done well (e.g., one just uses the default regularization parameter), DR does have relatively better performance—but these approaches result in lower performance overall. These latter results provide an explanation for why practitioners may be continuing to use DR without undertaking the necessary comparison to using the original features. However, this practice seems generally wrongheaded in light of the main results, if the goal is to maximize generalization performance.

EMI Radiation Prediction and Structure Optimization of Packages by Deep Learning

With a rapid increase in operating frequency and package complexity, conventional analysis methods cannot efficiently cope with current complex electromagnetic interference (EMI) issues. In this paper, a new method built on a deep neural network (DNN) model is proposed to accurately and rapidly predict the maximum radiated electric field at 3-meters of a wire-bond ball grid array (WB-BGA) package. The key hyper-parameters of the DNN model, such as the learning rate and type of optimizers, are discussed in depth so as to attain optimal performance. Predicted radiation results by the DNN model and the results of the full-wave simulation show good agreement. Once DNN is trained, the prediction time is in the order of milliseconds and the model size is in megabytes, which can acquire the predicted radiation quickly and accurately and save storage space. Furthermore, to prevent the radiation from exceeding requirements, package structures are optimized by adjusting those parameters sensitive to radiation and disregarding insensitive parameters. The sensitivity of these WB-BGA package structural parameters to EMI radiation can be analyzed quickly based on data with different deviations generated by the trained DNN model. The sensitive parameters are adjusted according to their correlation with EMI radiation. The effectiveness and feasibility of the optimization method are verified by the WB-BGA package with two sets of different structural parameters.

An Integrated Approach to Empirical Bayesian Whole Genome Prediction Modeling

Computational efficiency is an increasing concern for whole genome prediction (WGP) based on denser genetic marker panels such that algorithms other than Markov Chain Monte Carlo (MCMC) warrant greater consideration, particularly for hierarchical models that flexibly confer either heavy-tailed (e.g., BayesA) or stochastic search and variable selection (SSVS) instead of Gaussian specifications on marker effect distributions. The expectation maximization (EM) algorithm is one attractive alternative; however, recently proposed hierarchical model implementations of EM have not addressed formal estimation of underlying hyperparameters even though their specifications are known to impact WGP accuracy. Furthermore, EM can be sensitive to starting values. We develop and explore the properties of an empirical Bayes strategy by conditioning EM implementations of BayesA or SSVS WGP models on marginal modal estimation of variance components and other key hyperparameters. These empirical Bayes implementations are compared against their MCMC counterparts for estimation of hyperparameters and WGP accuracy, both within the context of a simulation study and application to a loblolly pine dataset. In all cases, starting values were deemed to be important for EM-based estimates. Starting values based on MCMC posterior means were preferable, whereas those based on setting all marker effects equal to zero generally led to inferior performance. Nevertheless, a recently proposed regularization procedure was useful in alleviating the impact of starting values in the EM implementation of the SSVS model, as was modifying the expectation step in the BayesA model to be based on relative variances rather than on relative precisions.