ML-MIDTERM-HANDOUT.pdf
Document Details
Uploaded by UndisputableTechnetium
Saint Louis University
Tags
Related
- CS488-Ch-9A Machine Learning PDF
- Artificial Intelligence with Machine Learning (Unit 1) PDF
- Unit 1 - Introduction to Machine Learning PDF
- Artificial Intelligence PDF
- DST301 Artificial Intelligence Applications Lecture 02 - Machine Learning PDF
- CPCS-335 Introduction to Artificial Intelligence Lecture 8 PDF
Full Transcript
IT ELECTIVE-3: MACHINE LEARNING MIDTERM HANDOUT Types of Machine Learning Supervised Learning: Uses labeled data to train models. Goal of Supervised Learning Primary Goal: Predicting output from labeled input data. Classification Algorithms Com...
IT ELECTIVE-3: MACHINE LEARNING MIDTERM HANDOUT Types of Machine Learning Supervised Learning: Uses labeled data to train models. Goal of Supervised Learning Primary Goal: Predicting output from labeled input data. Classification Algorithms Common Algorithm: Logistic Regression. K-Means Clustering Usage Typical Scenario: Segmenting customers into groups. Purpose of Cross-Validation Model Evaluation: Ensures the model generalizes well to unseen data. Epoch in Neural Networks Definition: A single forward and backward pass of all the training examples. Classification Model Metrics Not a Metric: Mean Squared Error. Overfitting in Machine Learning Explanation: Model performs well on training data but poorly on test data. Role of Activation Function Purpose: To introduce non-linearity. Feature Selection Method Example: Principal Component Analysis. Regression Algorithms Common Algorithm: Linear Regression. Confusion Matrix Usage Purpose: To evaluate the performance of a classification algorithm. Sequential Decision Making Approach: Reinforcement Learning. Normalization in Data Preprocessing Purpose: To scale data to a standard range. Benefits of Deep Learning Models Advantage: They automatically learn feature representations. Convolutional Neural Networks (CNNs) Use Case Common Use Case: Image recognition. Hyperparameter in Machine Learning Definition: A parameter that is set before the learning process begins. Ensemble Learning Explanation: Using multiple models to achieve better performance. Unsupervised Learning Example Example: K-Means Clustering. Curse of Dimensionality DR. BENEDICT SY IT ELECTIVE-3: MACHINE LEARNING MIDTERM HANDOUT Definition: The problem of overfitting in high-dimensional spaces. Policy in Reinforcement Learning Role: The strategy that the agent employs to determine its actions. Imbalanced Classification Metric Appropriate Metric: Precision-Recall Curve. Backpropagation Algorithm Purpose Function: To update the weights by minimizing the loss function. False Positive in Classification Definition: Incorrectly identifying a negative class as positive. Handling Missing Data Technique: Imputation. k-Nearest Neighbors (k-NN) Algorithm Characteristic: It uses a distance metric to classify new points. Bootstrapping in Machine Learning Definition: A resampling technique used to estimate statistics on a population. Preventing Overfitting Method: Applying regularization techniques. Dropout in Neural Networks Purpose: A regularization method to prevent overfitting. Principal Component Analysis (PCA) Goal: To transform data into a lower-dimensional space. DR. BENEDICT SY