2025 · 31 citations · 7 references
ComputingBiomedical SciencesComputer ScienceData MiningMachine LearningStatistical LearningClassificationDecision TreesEnsemble LearningOptimization For Machine LearningProbabilistic Machine LearningSupport Vector MachinesBenchmark Dataset EvaluationBreast Tumor ClassificationClassification AccuracyDecision Tree ClassificationHyperparameter OptimizationHyperparameter TuningLogistic RegressionRandom Forest Classification
Accurate classification of breast tissue is important for the early medical conditions like carcinoma, fibro-adenoma, and mastopathy. Machine learning models hold promise for automating it, but their performance significantly depends on the selection of the optimal hyperparameters. Traditional optimization techniques, such as grid search or random search, are inefficient and create too much computational cost in handling high-dimensional parameter spaces. In this paper, we present the use of Tree-Structured Parzen Estimator (TPE), a Bayesian probabilistic optimization technique, to optimize hyperparameters for five already wellknown machine learning classifiers: Logistic Regression, SVM, Decision Tree, Random Forest, and k-NN. We adopt this approach to the Breast Tissue Dataset as our benchmark dataset to classify various types of breast tissues. We assess our models with respect to four key performance metrics: Accuracy, Precision, Recall, and F1-Score. The experiments show that TPE optimization leads to consistent improvements in classification accuracy for all the models. For precision, recall, and F1-score, significant gains were observed. Besides, TPE improves model generalization by focusing the search on promising regions of the hyperparameter space, reducing the computational cost compared to traditional methods. This work shows the potential of TPE as a powerful tool for hyperparameter optimization in medical machine learning tasks, providing a more efficient and effective alternative to conventional optimization strategies.
7
LightGBM: A Highly Efficient Gradient Boosting Decision Tree
Guolin Ke, Qi Meng, Thomas Finley et al. · HAL (Le Centre pour la Communication Scientifique Directe) · 2017 · 9.5K citations · Full text
Random search for hyper-parameter optimization
James Bergstra, Yoshua Bengio · 2012 · 7.9K citations