Comprehensive Final Exam Review: Data Structures, Similarity Measures, and Machine Learning Fundamentals
Focus on the following key concepts and subtopics for the upcoming exam:
Kaomi Sketch and Bloom Filter Comparisons
Purpose and functionality of the Kaomi sketch vs. Bloom filter
Understanding error bounds and configuration parameters
Role of the two-dimensional array in mapping and storing data
Jaccard and Cosine Similarity Measures
Computation methods for Jaccard and Cosine similarity
Differences between similarity and distance metrics
Importance of the bag of words model in text analysis
Mathematical Norms and Manhattan Distance
Familiarity with various norms
Relationship between L1 norm and Manhattan distance
KD Trees in Nearest Neighbor Search
Construction and functioning of KD trees
Nearest neighbor search processes and back checking
Curse of Dimensionality and Feature Reduction
Understanding overfitting and its relation to dimensionality
Methods of dimensionality reduction, including PCA
MinHash and Similarity Computation
Relationship between the MinHash algorithm and Jaccard similarity
Random hash functions and their applications
Generalization, Fitting, and Bias-Variance
Definitions and differences of key terms
Examples of bias and variance relationships
KL Divergence Formulas
Definition and calculation processes
Confusion Matrix and Evaluation Metrics
Calculation of True Positives, False Positives, etc.
Relationships among accuracy, precision, recall, and F1 score
ROC Curves and AUC Analysis
Understanding ROC curve utility, TPR, FPR, and specificity
AUC calculation and classifier comparison
Linear Regression Construction and Optimization
Construction of linear models and understanding cost functions
Logistic Regression and Maximum Likelihood
Writing optimization objectives of MLE
Regularization: Empirical vs. Structured Risk
Differences between ERM and SRM
Principal Component Analysis (PCA)
Steps of PCA and application in feature reduction
Singular Value Decomposition (SVD)
Purpose and steps involved in SVD implementation.