China 
Africa 
Explore chapters and articles related to this topic
Introduction to Deep Learning
Published in Lia Morra, Silvia Delsanto, Loredana Correale, Artificial Intelligence in Medical Imaging, 2019
Lia Morra, Silvia Delsanto, Loredana Correale
Manual intuition-based search is arguably the most commonly used technique to optimize the model and hyper-parameters, because it is easy to implement and exploits researchers’ experience to reduce the number of trials, which is useful when training large networks that require extensive computational resources [28]. Nonetheless, in recent years there has been increasing interest in meta-learning or self-supervising strategies that can automate the process of model selection and training, which goes under the name of Automated Machine Learning or AutoML.
Introduction
Published in Chandrasekar Vuppalapati, Democratization of Artificial Intelligence for the Future of Humanity, 2021
AutoML: Automated Machine Learning is the capability of automating buildings, hyper-tuning, deploying, and managing ML models. With AutoML capabilities, data scientists could compare the accuracies of various ML models and pivot the model selection to the most accurate technique in an automated and time efficient manner. Microsoft Auto ML,79 Cloud AutoML80 from Google, Amazon Sagemaker,81 DataRobot82 Automated Machine Learning and H2O83—Automated ML.
Deep Learning
Published in Peter Wlodarczak, Machine Learning and its Applications, 2019
Feature learning is part of an effort to automate the whole end-to-end machine learning process, called automated machine learning. Automated machine learning aims to automate all machine learning tasks that are usually done manually, such as model selection or hyperparameter. However, automatic feature extraction is not always feasible. Feature learning is not restricted to deep learning and has been used with shallow learners, such as k-means clustering or principal component analysis.
Comparative Study of AutoML Approach, Conventional Ensemble Learning Method, and KNearest Oracle-AutoML Model for Predicting Student Dropouts in Sub-Saharan African Countries
Published in Applied Artificial Intelligence, 2022
Yuda N Mnyawami, Hellen H Maziku, Joseph C Mushi
Automated Machine Learning (AutoML) is a technique to drive the best classification model and corresponding hyper-parameter for a given decision-making problem (Feurer et al. 2015). The AutoML selects the best combination of hyper-parameters and features for the optimal prediction model (Kotthof et al. 2017). The process of building such actionable machine learning models can generate added value to the existing problem (Tuggener et al. 2019). Several studies have predicted student performance and dropout rate in higher education using traditional machine learning models, and a few studies have used AutoML models (Zeineddine, Braendle, and Farah 2021). The KNORA-AutoML method for improving prediction models for student dropout is not yet covered in secondary schools in developing countries. Prabaharan, Mehta, and Chauhan (2020) and Waring, Lindvall, and Umeton (2020) evidenced the significance of using the AutoML method in healthcare, demonstrating the influential features and the best combination of the classification methods to improve health outcomes. Krauß et al. (2020) applied the AutoML model using hyper-parameter optimization techniques in quality production. Their preliminary results showed F1 scores of 37% for the un-tuned model and 42% for the tuned model using Random forest. Automatic generation of influential features and machine learning algorithms lead to the optimal prediction model implemented by hyper-parameter optimization techniques (Feurer et al. 2015).
Quantifying alignment deviations for uniaxial material mechanical testing via automated machine learning
Published in Mechanics of Advanced Materials and Structures, 2022
Junxian Chen, Jianhai Zhang, Hongwei Zhao
The Automated machine learning (AutoML) aims to ease the demand for data scientists by automatically assisting domain experts in constructing machine learning (ML) applications without the requirement of extensive statistical or ML knowledge [23]. The AutoML enables ML experts to automate mundane tasks such as hyperparameter optimization, leading to better performance. In general, it is essential to construct an artificial neural network to clean the data, extract features, and optimize the super parameters. For instance, Moslemi et al. predicted the subsequent heat and glass transition behavior [24] and Goldak welding simulation parameters [25]. However, mature automatic machine learning toolkits, such as Tpot, H2O, autoweka, auto sklearn, autogluon, and Google automl tables, can automatically realize most of the processes in the artificial neural network, and yield an efficient and high-precision machine learning model by employing a variety of optimization strategies in the solution procedure. Domain specialists can be empowered to design machine-learning pipelines independently of data scientists. Currently, AutoML is devoted entirely to supervised learning. Researchers should constantly evaluate their proposed approaches for supervised learning, even if some methods have been utilized for unsupervised or reinforcement learning. Unsupervised or reinforcement learning research may accelerate the development of the AutoML framework for previously undiscovered learning challenges. Furthermore, specialized strategies may enhance the performance of specific activities.
Credit Card Fraud Detection with Automated Machine Learning Systems
Published in Applied Artificial Intelligence, 2022
Vasilios Plakandaras, Periklis Gogas, Theophilos Papadimitriou, Ioannis Tsamardinos
To respond to such challenges, systems and services that automate a large part of the machine learning pipeline have recently appeared under the name of Automated Machine Learning (AutoML) system. Such systems automate the selection of ML algorithms, the tuning of their hyper-parameter values, the estimation of performance, and the visualization and interpretation of results. In this paper, we demonstrate how AutoML tools could potentially increase the productivity of detecting fraudulent credit card transactions without a reduction in the prediction performance compared to a manual analysis. Specifically, we describe and use the Just Add Data Bio1 (hereafter JAD) AutoML tool on the fraud detection problem described above and achieve results on par with state-of-the-art previous analyses that are manually coded. Secondly, in addition to modeling, JAD performs automated feature selection to identify the most significant variables to fraud detection, providing valuable intuition to fraud inspectors. We’d like to note that JAD’s feature selection considers features jointly (multivariate) and not simply one by one. Features that are informative by themselves may become redundant given other features; similarly, features that are uninformative by themselves may be necessary for optimal prediction and become informative given other features. Hence, optimal feature selection is a combinatorial problem that returns the minimal-size feature subset that in combination leads to the optimally predictive model. After examining numerous combinations of algorithms for feature selection and modeling, as well as their tuning hyper-parameter values, JAD selects the best one to create a final model for prediction. It estimates its predictive performance along several common metrics (e.g., AUC, accuracy, balanced accuracy, F1 score), the confidence intervals of performance, the Receiver Operating Characteristic (ROC) curve, and the contribution to performance for each selected feature.