Hyperparameter Optimization Techniques Research Articles

Spatio-temporal air quality analysis and PM2.5 prediction over Hyderabad City, India using artificial intelligence techniques

Air pollution is one of the most serious environmental issues faced by humans, and it affects the quality of life in cities. PM2.5 forecasting models can be used to create strategies for assessing and warning the public about anticipated harmful levels of air pollution. Accurate pollutant concentration measurements and forecasting are critical criteria for assessing air quality and are the foundation for making the right strategic decisions. Data-driven machine learning models for PM2.5 forecasting have gained attention in the recent past. In this study, PM2.5 prediction for Hyderabad city was carried out using various machine learning models viz. Multi-Linear Regression (MLR), decision tree (DT), K-Nearest Neighbors (KNN), Random Forest (RF), and XGBoost. A deep learning model, the Long Short-Term Memory (LSTM) model, was also used in this study. The results obtained were finally compared based on error and R2 value. The best model was selected based on its maximum R2 value and minimal error. The model's performance was further improved using the randomized search CV hyperparameter optimization technique. Spatio-temporal air quality analysis was initially conducted, and it was found that the average winter PM2.5 concentrations were 68% higher than the concentrations in summer. The analysis revealed that XGBoost regression was the best-performing machine learning model with an R2 value of 0.82 and a Mean Absolute Error (MAE) of 7.01 μg/ m3, whereas the LSTM deep learning model performed better than XGBoost regression for PM2.5 modeling with an R2 value of 0.89 and an MAE of 5.78 μg/ m3.

Feature extraction and prediction of fine particulate matter (PM2.5) chemical constituents using four machine learning models

The concentrations of fine particulate matter (PM2.5) constituents, which are very important and essential information for the identification of air pollution sources, were predicted at three sites (Seoul, Ulsan, Baengnyeong) in South Korea between 2016 and 2018 using four machine learning (ML) models: generative adversarial imputation network (GAIN), fully connected deep neural network (FCDNN), random forest (RF), and k-nearest neighbor (kNN). 3 PM2.5 constituent groups, namely 8 ions, 2 carbons, and 15 trace elements, were targeted for prediction. The latest hyperparameter optimization techniques were used to learn air pollution characteristics from ambient PM2.5-related information, such as time, meteorology, and air pollutant concentrations. We compared the feature extraction abilities of the four models. The prediction accuracy identified by the coefficient of determination (R2) between prediction and observation was highest in GAIN, followed by FCDNN and RF or kNN. On availability of data on the time, air pollutant concentrations, and/or meteorology, simultaneously missed 20 % data of all PM2.5 constituent groups were predicted, with R2 = 0.897, 0.861, 0.785, and 0.801 by the GAIN, FCDNN, RF, and kNN, respectively. As missing ratios (20 %, 40 %, 60 %, 80 %) of input data increased, prediction accuracy decreased in the four models and was predominantly more noticeable in GAIN and kNN. As the available period of data increased, prediction accuracy increased in GAIN and FCDNN. Trace elements were predicted with the lowest R2 in all models among the target constituent groups. Study sites with more emission sources showed lower prediction accuracy, resulting in the highest R2 in Baengnyeong island and the lowest in Ulsan. According to the current findings, ML models can be used to evaluate various air pollution issues for which data is missing.

Prediction of compressive strength of recycled aggregate concrete using machine learning and Bayesian optimization methods

With the sustainable development of the construction industry, recycled aggregate (RA) has been widely used in concrete preparation to reduce the environmental impact of construction waste. Compressive strength is an essential measure of the performance of recycled aggregate concrete (RAC). In order to understand the correspondence between relevant factors and the compressive strength of recycled concrete and accurately predict the compressive strength of RAC, this paper establishes a model for predicting the compressive strength of RAC using machine learning and hyperparameter optimization techniques. RAC experimental data from published literature as the dataset, extreme gradient boosting (XGBoost), random forest (RF), K-nearest neighbour (KNN), support vector machine regression Support Vector Regression (SVR), and gradient boosted decision tree (GBDT) RAC compressive strength prediction models were developed. The models were validated and compared using correlation coefficients (R2), Root Mean Square Error (RMSE), mean absolute error (MAE), and the gap between the experimental results of the predicted outcomes. In particular, The effects of different hyperparameter optimization techniques (Grid search, Random search, Bayesian optimization-Tree-structured Parzen Estimator, Bayesian optimization- Gaussian Process Regression) on model prediction efficiency and prediction accuracy were investigated. The results show that the optimal combination of hyperparameters can be searched in the shortest time using the Bayesian optimization algorithm based on TPE (Tree-structured Parzen Estimator); the BO-TPE-GBDT RAC compressive strength prediction model has higher prediction accuracy and generalisation ability. This high-performance compressive strength prediction model provides a basis for RAC’s research and practice and a new way to predict the performance of RAC.

A Transfer Learning and Optimized CNN Based Maritime Vessel Classification System

Deep learning has been used to improve intelligent transportation systems (ITS) by classifying ship targets in interior waterways. Researchers have created numerous classification methods, but they have low accuracy and misclassify other ship targets. As a result, more research into ship classification is required to avoid inland waterway collisions. We present a new convolutional neural network classification method for inland waterways that can classify the five major ship types: cargo, military, carrier, cruise, and tanker. This method can also be used for other ship classes. The proposed method consists of four phases for the boosting of classification accuracy for Intelligent Transport Systems (ITS) based on convolutional neural networks (CNNs); efficient augmentation method, the hyper-parameter optimization (HPO) technique for optimum CNN model parameter selection, transfer learning, and ensemble learning are suggested. All experiments used Kaggle’s public Game of Deep Learning Ship dataset. In addition, the proposed ship classification achieved 98.38% detection rates and 97.43% F1 scores. Our suggested classification technique was also evaluated on the MARVEL dataset. This dataset includes 10,000 image samples for each class and 26 types of ships for generalization. The suggested method also delivered an excellent performance compared to other algorithms, with performance metrics with an accuracy of 97.04%, a precision of 96.1%, a recall of 95.92%, a specificity of 96.55%, and a 96.31% F1 score.

Internal Leakage Detection in Hydraulic Pump Using Model-Agnostic Feature Ranking and Ensemble Classifiers

Abstract Hydraulic pumps are key drivers of fluid power-based machines and demand high reliability during operation. Internal leakage is a key performance deteriorating fault that reduces pump’s efficiency and limits its predictability and reliability. Thus, this article presents a methodology for detecting internal leakage in hydraulic pumps using an unbalanced dataset of its drive motor’s electrical power signals. Refined composite multiscale dispersion and fuzzy entropies along with three statistical indicators are extracted and followed by second-order polynomial-based features. These features are normalized and visualized using partial dependence plot (PDP) and individual conditional expectation (ICE). Subsequently, ten machine learning classifiers are trained using four features, and their statistical hypothesis test is performed using a 5 × 2 paired t-test cross-validation for p < 0.05. Subsequently, top four performing classifiers are optimized using grid and random search hyperparameter optimization techniques. Due to slight difference in their accuracies, an ensemble of three best-performing algorithms is trained using the majority voting classifiers (MaVCs) for three splitting ratios (80:20, 70:30, and 60:40). It is demonstrated that MaVC achieves the highest leakage detection accuracy of 90.91%.

Prediction modelling framework comparative analysis of dissolved oxygen concentration variations using support vector regression coupled with multiple feature engineering and optimization methods: A case study in China

Dissolved oxygen (DO) is an essential indicator for assessing water quality and managing aquatic environments, but it is still a challenging topic to accurately understand and predict the spatiotemporal variation of DO concentrations under the complex effects of different environmental factors. In this study, a practical prediction framework was proposed for DO concentrations based on the support vector regression (SVR) model coupling multiple intelligence techniques (i.e., four data denoising techniques, three feature selection rules, and four hyperparameter optimization methods). The holistic framework was tested using a data matrix (17,532 observation data in total) of 12 indicators from three vital water quality monitoring stations of the longest inter-basin water diversion project in the world (i.e., the Middle-Route of the South-to-North Water Diversion Project of China), during the year 2017 to 2020 period. The results showed that the framework we advocated for could successfully and accurately predict DO concentration variations in different geographical locations. The model used the “wavelet analysis–LASSO regression–random search–SVR” combination of the Waihuanhe station has the best prediction performance, with the Root Mean Square Error (RMSE), Mean Square Error (MSE), Mean Absolute Error (MAE), and coefficient of determination (R2) values of 0.251, 0.063, 0.190, and 0.911, respectively. The combined methods using feature selection and hyperparameter optimization techniques can significantly promote the robustness and accuracy of the prediction model and can provide a new universal and practical way of investigating and understanding the environmental drivers of DO concentration variations. For the water quality management department, this proposed comprehensive framework can also identify and reveal the key parameters that should be concerned and monitored under different environmental factors change. More studies in terms of assessing potential integrated water quality risk using multi-indicators in mega water diversion projects and/or similar water bodies are required in the future.

Automated hyperparameter optimization for simulating boundary conditions for acoustic and elastic wave propagation using deep learning

We have developed a deep learning framework to simulate the effect of boundary conditions for wave propagation in anisotropic media. To overcome the challenges associated with the stability of conventional implementation of boundary conditions for strongly anisotropic media, we develop an efficient algorithm using deep neural networks. We incorporate the hyperparameter optimization (HPO) workflow in the deep learning framework to automate the network training process. Hyperparameter selection is a crucial step in model building and has a direct impact on the performance of machine learning models. We implement three different HPO techniques, namely random search, Hyperband, and Bayesian optimization, for simulating boundary conditions and compare the strengths and drawbacks of these techniques. We train the network using a few shot locations and time slices enabling the network to learn how to remove boundary reflections and simulate wave propagation for unbounded media. The automated deep learning framework with HPO improves the performance of deep learning models by achieving the optimal minima and significantly improves the efficiency of the workflow. The benefit of this approach is its simple implementation and significant reduction of reflections at the boundaries, especially in the case of tilted transverse isotropic media. We validate our approach by comparing wave propagation at the boundaries using our algorithm with the output obtained using the unbounded media simulated by padding the model. Tests on different models with acoustic and elastic wave propagation verify the effectiveness of our approach.

Automatic Detection of Software Defects based on Machine Learning

Defects in software are one of the critical problems in software engineering community because they provide inaccurate results and negatively affect the quality and reliability of the software. These defects must be detected in the early stages of software development. Researchers had used Software Defect Detection (SDD) techniques to allow predicting module fault-proneness. By implementing the hyperparameter optimization techniques and exploiting data imbalances in predicting defects, this paper proposes and develops an SDD model with high performance and generalization capability. To classify defects in software modules, machine learning algorithms and ensemble learning techniques are used on the balanced datasets. The balanced datasets are obtained through using a hybrid of synthetic minority oversample (SMOTE) and Support Vector Machine (SVM). To obtain the optimal hyperparameters needed for the used classifiers and for the dataset balanced algorithms, Non-dominated Sorting Genetic Algorithm II (NDSGA-II) is used. To reduce the time and save other used resources, Hyperband technique, which is a multi-fidelity optimization, is used in NDSGA-II. A 10-fold Cross Validation (CV) is applied to overcome the overfitting and underfitting problems. The accuracy, recall, F-measure, and ROC AUC metrics are used to evaluate the SDD model. The results show that the proposed model predicts defects more accurately than the compared studies.

The Role of Machine Learning in Identifying Students At-Risk and Minimizing Failure

Education is very important for students’ future success. The performance of students can be supported by the extra assignments and projects given by the instructors for students with low performance. However, a major problem is that students at-risk cannot be identified early. This situation is being investigated by various researchers using Machine Learning techniques. Machine learning is used in a variety of areas and has also begun to be used to identify students at-risk early and to provide support by instructors. This research paper discusses the performance results found using Machine learning algorithms to identify at-risk students and minimize student failure. The main purpose of this project is to create a hybrid model using the ensemble stacking method and to predict at-risk students using this model. We used machine learning algorithms such as Naïve Bayes, Random Forest, Decision Tree, K-Nearest Neighbors, Support Vector Machine, AdaBoost Classifier and Logistic Regression in this project. The performance of each machine learning algorithm presented in the project was measured with various metrics. Thus, the hybrid model by combining algorithms that give the best prediction results is presented in this study. The data set containing the demographic and academic information of the students was used to train and test the model. In addition, a web application developed for the effective use of the hybrid model and for obtaining prediction results is presented in the report. In the proposed method, it has been realized that stratified k-fold cross validation and hyperparameter optimization techniques increased the performance of the models. The hybrid ensemble model was tested with a combination of two different datasets to understand the importance of the data features. In first combination, the accuracy of the hybrid model was obtained as 94.8% by using both demographic and academic data. In the second combination, when only academic data was used, the accuracy of the hybrid model increased to 98.4%. This study focuses on predicting the performance of at-risk students early. Thus, teachers will be able to provide extra assistance to students with low performance.

EISM-CPS: An Enhanced Intelligent Security Methodology for Cyber-Physical Systems through Hyper-Parameter Optimization

The increased usage of cyber-physical systems (CPS) has gained the focus of cybercriminals, particularly with the involvement of the internet, provoking an increased attack surface. The increased usage of these systems generates heavy data flows, which must be analyzed to ensure security. In particular, machine learning (ML) and deep learning (DL) algorithms have shown feasibility and promising results to fulfill the security requirement through the adoption of intelligence. However, the performance of these models strongly depends on the model structure, hyper-parameters, dataset, and application. So, the developers only possess control over defining the model structure and its hyper-parameters for diversified applications. Generally, not all models perform well in default hyper-parameter settings. Their specification is a challenging and complex task and requires significant expertise. This problem can be mitigated by utilizing hyper-parameter optimization (HPO) techniques, which intend to automatically find efficient learning model hyper-parameters in specific applications or datasets. This paper proposes an enhanced intelligent security mechanism for CPS by utilizing HPO. Specifically, exhaustive HPO techniques have been considered for performance evaluation and evaluation of computational requirements to analyze their capabilities to build an effective intelligent security model to cope with security infringements in CPS. Moreover, we analyze the capabilities of various HPO techniques, normalization, and feature selection. To ensure the HPO, we evaluated the effectiveness of a DL-based artificial neural network (ANN) on a standard CPS dataset under manual hyper-parameter settings and exhaustive HPO techniques, such as random search, directed grid search, and Bayesian optimization. We utilized the min-max algorithm for normalization and SelectKBest for feature selection. The HPO techniques performed better than the manual hyper-parameter settings. They achieved an accuracy, precision, recall, and F1 score of more than 98%. The results highlight the importance of HPO for performance enhancement and reduction of computational requirements, human efforts, and expertise.

Automatic classification of Candida species using Raman spectroscopy and machine learning

One of the problems that most affect hospitals is infections by pathogenic microorganisms. Rapid identification and adequate, timely treatment can avoid fatal consequences and the development of antibiotic resistance, so it is crucial to use fast, reliable, and not too laborious techniques to obtain quick results. Raman spectroscopy has proven to be a powerful tool for molecular analysis, meeting these requirements better than traditional techniques. In this work, we have used Raman spectroscopy combined with machine learning algorithms to explore the automatic identification of eleven species of the genus Candida, the most common cause of fungal infections worldwide. The Raman spectra were obtained from more than 220 different measurements of dried drops from pure cultures of each Candida species using a Raman Confocal Microscope with a 532 nm laser excitation source. After developing a spectral preprocessing methodology, a study of the quality and variability of the measured spectra at the isolate and species level, and the spectral features contributing to inter-class variations, showed the potential to discriminate between those pathogenic yeasts. Several machine learning and deep learning algorithms were trained using hyperparameter optimization techniques to find the best possible classifier for this spectral data, in terms of accuracy and lowest possible overfitting. We found that a one-dimensional Convolutional Neural Network (1-D CNN) could achieve above 80 % overall accuracy for the eleven classes spectral dataset, with good generalization capabilities.

Modeling of Optimal Fully Connected Deep Neural Network based Sentiment Analysis on Social Networking Data

Abstract In recent times, the user reviews and opinions postedon social-media like Twitter, Facebook, and Google+, and so on have received considerable attention in the domain of sentiment analysis (SA), offering valued feedback to public as well as private organizations. The examination of reviews roles an essential role for enhancing the product and service qualities. Although numerous SA methods are available in the literature, it is still needed to boost the outcome of the SA for understanding the customer feedbacks and thereby enhances the product quality. The presented article proposesan optimum fully connected deep neural network (OFCDNN) based SA, called OFCDNN-SA technique on social networking data. The OFCDNN-SA technique encompasses distinct procedureslike feature extraction, classification, preprocessing, and parameter optimization. Additionally, Glove technique is applied for the transformation of input data into feature vectors. Moreover, salp swarm optimization (SSO) based hyperparameter optimization technique is derived for maximally selecting the hyperparameters utilized in the DL method. Finally, the FCDNN methodis employed for classification purposes. To investigate the supreme achievement of the OFCDNN-SA method, a wide-ranging simulation assessment is performed and the experimental outputs highlighted the betterment over current methods by means of diverse measures.

A boosting ensemble learning based hybrid light gradient boosting machine and extreme gradient boosting model for predicting house prices

AbstractThe implementation of tree‐ensemble models has become increasingly essential in solving classification and prediction problems. Boosting ensemble techniques have been widely used as individual machine learning algorithms in predicting house prices. One of the techniques is LGBM algorithm that employs leaf wise growth strategy, reduces loss and improves accuracy during training which results in overfitting. However, XGBoost algorithm uses level wise growth strategy which takes time to compute resulting in higher computation time. Nevertheless, XGBoost has a regularization parameter, implements column sampling and weight reduction on new trees which combats overfitting. This study focuses on developing a hybrid LGBM and XGBoost model in order to prevent overfitting through minimizing variance whilst improving accuracy. Bayesian hyperparameter optimization technique is implemented on the base learners in order to find the best combination of hyperparameters. This resulted in reduced variance (overfitting) in the hybrid model since the regularization parameter values were optimized. The hybrid model is compared to LGBM, XGBoost, Adaboost and GBM algorithms to evaluate its performance in giving accurate house price predictions using MSE, MAE and MAPE evaluation metrics. The hybrid LGBM and XGBoost model outperformed the other models with MSE, MAE and MAPE of 0.193, 0.285, and 0.156 respectively.

Classification of SSVEP-EEG signals using CNN and Red Fox Optimization for BCI applications.

Classification of electroencephalography (EEG) signals associated with Steady-state visually evoked potential (SSVEP) is prominent because of its potential in restoring the communication and controlling capability of paralytic people. However, SSVEP signals classification is a challenging task for researchers because of its low signal-to-noise ratio, non-stationary and high dimensional properties. A proficient technique has to be evolved to classify the SSVEP-based EEG data. In recent times, convolutional neural network (CNN) has reached a quantum leap in EEG signal classification. Therefore, the proposed system employs CNN to classify the SSVEP-based EEG signals. Though CNN has proved its proficiency in handling EEG signal classification problems, the calibration of hyperparameters is required to enhance the performance of the model. The calibration of a hyperparameter is a time-consuming task, hence proposed an automated hyperparameter optimization technique using the Red Fox Optimization Algorithm (RFO). The effectiveness of the algorithm is evaluated by comparing it with the performance of Harris Hawk Optimization (HHO), Flower Pollination Algorithm (FPA), Grey Wolf Optimization Algorithm (GWO) and Whale Optimization Algorithm (WOA) based hyperparameter optimized CNN applied to the SSVEP based EEG signals multiclass dataset. The experimental results infer that the proposed algorithm can achieve a testing accuracy of 88.91% which is higher than other comparative algorithms like HHO, FPA, GWO and WOA. The above-mentioned values clearly show that the proposed algorithm achieved competitive performance when compared to the other reported algorithm.

Improved Fine-Tuning of In-Domain Transformer Model for Inferring COVID-19 Presence in Multi-Institutional Radiology Reports.

Building a document-level classifier for COVID-19 on radiology reports could help assist providers in their daily clinical routine, as well as create large numbers of labels for computer vision models. We have developed such a classifier by fine-tuning a BERT-like model initialized from RadBERT, its continuous pre-training on radiology reports that can be used on all radiology-related tasks. RadBERT outperforms all biomedical pre-trainings on this COVID-19 task (P<0.01) and helps our fine-tuned model achieve an 88.9 macro-averaged F1-score, when evaluated on both X-ray and CT reports. To build this model, we rely on a multi-institutional dataset re-sampled and enriched with concurrent lung diseases, helping the model to resist to distribution shifts. In addition, we explore a variety of fine-tuning and hyperparameter optimization techniques that accelerate fine-tuning convergence, stabilize performance, and improve accuracy, especially when data or computational resources are limited. Finally, we provide a set of visualization tools and explainability methods to better understand the performance of the model, and support its practical use in the clinical setting. Our approach offers a ready-to-use COVID-19 classifier and can be applied similarly to other radiology report classification tasks.

Segmentation of tomography datasets using 3D convolutional neural networks

Dendritic microstructures are ubiquitous in nature and are the primary solidification morphologies in metallic materials. Techniques such as X-ray computed tomography (XCT) have provided new insights into dendritic phase transformation phenomena. However, manual identification of dendritic morphologies in microscopy data can be both labor intensive and potentially ambiguous. The analysis of 3D datasets is particularly challenging due to their large sizes (terabytes) and the presence of artifacts scattered within the imaged volumes. In this study, we trained 3D convolutional neural networks (CNNs) to segment 3D datasets. Three CNN architectures were investigated, including a new version of FCDenseNet which we extended to 3D. We show that using hyperparameter optimization (HPO) and fine-tuning techniques, both 2D and 3D CNN architectures outperform the previous state of the art. The 3D U-Net architecture trained in this study produced the best segmentations according to quantitative metrics (intersection-over-union of 95.56% and a boundary displacement error of 0.58 pixels), while 3D FCDense produced the smoothest boundaries and best segmentations according to visual inspection. The trained 3D CNNs are able to segment entire 852 × 852 × 250 voxel 3D volumes in only ∼60 s, thus hastening the progress towards a deeper understanding of phase transformation phenomena such as dendritic solidification.

DEHypGpOls: a genetic programming with evolutionary hyperparameter optimization and its application for stock market trend prediction

Stock markets are a popular kind of financial markets because of the possibility of bringing high revenues to their investors. To reduce risk factors for investors, intelligent and automated stock market forecast tools are developed by using computational intelligence techniques. This study presents a hyperparameter optimal genetic programming-based forecast model generation algorithm for a-day-ahead prediction of stock market index trends. To obtain an optimal forecast model from the modeling dataset, a differential evolution (DE) algorithm is employed to optimize hyperparameters of the genetic programming orthogonal least square (GpOls) algorithm. Thus, evolution of GpOls agents within the hyperparameter search space enables adaptation of the GpOls algorithm for the modeling dataset. This evolutionary hyperparameter optimization technique can enhance the data-driven modeling performance of the GpOls algorithm and allow the optimal autotuning of user-defined parameters. In the current study, the proposed DE-based hyper-GpOls (DEHypGpOls) algorithm is used to generate forecaster models for prediction of a-day-ahead trend prediction for the Istanbul Stock Exchange 100 (ISE100) and the Borsa Istanbul 100 (BIST100) indexes. In this experimental study, daily trend data from ISE100 and BIST100 and seven other international stock markets are used to generate a-day-ahead trend forecaster models. Experimental studies on 4 different time slots of stock market index datasets demonstrated that the forecast models of the DEHypGpOls algorithm could provide 57.87% average accuracy in buy–sell recommendations. The market investment simulations with these datasets showed that daily investments to the ISE100 and BIST100 indexes according to buy or sell signals of the forecast model of DEHypGpOls could provide 4.8% more average income compared to the average income of a long-term investment strategy.

Developing a latent space to describe full-sphere head-related transfer functions

Several prior studies have used machine-learning techniques, such as Principal Component Analysis (PCA), to investigate the information content in unidirectional Head-Related Transfer Funcions (HRTFs). This paper builds upon this prior work to investigate the information content in full-sphere, free-field HRTFs. An artificial neural network is trained to reproduce the PCA coefficients of an HRTF from the combination of a direction vector and latent space vector identifying the HRTF. The training data comes from several different HRTF databases; the methods used to combine these databases into a single coherent dataset are described. Hyperparameter optimization techniques are then used to minimize the dimensionality of the latent space vector, providing new information about the information content required to accruately describe a full-sphere HRTF.

Deep Learning-Based Classification of Spoken English Digits.

Classification of isolated digits is the basic challenge for many speech classification systems. While a lot of work has been carried out on spoken languages, only limited research work on spoken English digit data has been reported in the literature. The paper proposes an intelligent-based system based on deep feedforward neural network (DFNN) with hyperparameter optimization techniques, an ensemble method; random forest (RF), and a regression method; gradient boosting (GB) for the classification of spoken digit data. The paper investigates different machine learning (ML) algorithms to determine the best method for the classification of spoken English digit data. The DFNN classifier outperformed the RF and GB classifiers on the public benchmark spoken English digit data and achieved 99.65% validation accuracy. The outcome of the proposed model performs better compared to existing models with only traditional classifiers.

Benchmarking of hyperparameter optimization techniques for machine learning applications in production

Machine learning (ML) has become a key technology to leverage the potential of large data amounts that are generated in the context of digitized and connected production processes. In projects for developing ML solutions for production applications, the selection of hyperparameter optimization (HPO) techniques is a key task that significantly impacts the performance of the resulting ML solution. However, selecting the best suitable HPO technique for an ML use case is challenging, since HPO techniques have individual strengths and weaknesses and ML use cases in production are highly individual in terms of their application areas, objectives, and resources. This makes the selection of HPO techniques in production a very complex task that requires decision support. Thus, we present a structured approach for benchmarking HPO techniques and for integrating the empirical data generated within benchmarking experiments into decision support systems. Based on the data generated within a large-scale benchmarking study, the validation results prove that the usage of benchmarking data improves decision-making in HPO technique selection and thus helps to exploit the full potential of ML solutions in production applications.