지도 학습이란 레이블이 지정된 데이터를 사용해 모델을 훈련시켜 예측을 수행하는 머신러닝 기법의 한 유형입니다. 지도 학습에서 모델은 알려진 입력 데이터셋과 응답 변수를 사용하여 분류 또는 회귀 작업을 수행합니다.
Supervised learning is a type of machine learning that uses labeled data to train models to make predictions, where the algorithm learns from a known set of input data (features) paired with known responses or outputs.
The two main types are classification, which predicts categorical responses like whether equipment parts are defective or good, and regression, which predicts continuous values like electricity load or battery state of charge.
Supervised learning requires labeled training data with known inputs and outputs, while unsupervised learning uses unlabeled data to uncover hidden patterns without human intervention. Supervised learning results are typically more accurate but require time-consuming and potentially expensive data labeling.
Applications include medical imaging and tumor detection, ECG signal classification, electricity load forecasting, credit scoring, anomaly detection in industrial automation, and battery state-of-charge estimation in electric vehicles.
Decision trees, support vector machines (SVMs), and neural networks can be applied to both classification and regression tasks in supervised learning.
Model improvement involves feature engineering (including feature selection and transformation) and hyperparameter tuning to identify parameters like learning rate, batch size, and number of epochs that provide the best model performance.
MATLAB offers the Classification Learner and Regression Learner apps for training and evaluating machine learning models, the Deep Network Designer app for neural networks, and tools for data preprocessing, feature engineering, and model deployment to embedded systems.
Labeled data provides the correct input-output pairs that teach the model patterns and relationships, directly impacting prediction accuracy. The quality and realism of labeled data are crucial for model performance and generalization to unseen data.
자료
문서, 예제, 비디오 등을 통해 여러분의 지식을 확장할 수 있습니다.