🔬 Applied Scientist & ML Specialist938 Problems · 127 Topics🌐 Linked to Tech Vault

TalentMe Science Depth: Universal Applied Science & ML Depth Problem Bank

938 end-to-end science depth, system design, transition bridge & BQ problems covering derivations, production trade-offs, pitfalls, and follow-ups.

🎯

Science Depth Mastery Tracker

0% Mastered

Locally synced in your browser. Track, review, and master all 938 scientific & design topics.

Mastered: 0
Reviewing: 0
Unstarted: 119
🚀 Start Next Unpracticed →
🌐
Linked to TalentMe Tech Vault & MindmapsM2: Classical Machine Learning
All 938 questions are directly linked to TalentMe's Tech Vault guides, atomic cards, and interactive mindmaps.
🧭Classical ML Mindmap & Cards→📖Decision Trees, GBDT & XGBoost→🎯MLE Cheatsheet→
Found 24 in-depth questions
M2-025Classical Machine LearningDecision Trees
Medium

Explain the Bias-Variance Profile of Decision Trees and Why Individual Trees Suffer from High Variance

⚡ One-Sentence Answer

A fully grown decision tree has low bias but high variance; small perturbations in training data flip top-level splits, cascading down to alter the entire tree structure and producing erratic predictions.

Math & Tradeoffs→
M2-026Classical Machine LearningDecision Trees
Hard

Compare Industrial Strategies for Handling Missing Values and Categorical Features in Tree Models

⚡ One-Sentence Answer

Modern tree models (LightGBM, XGBoost) handle missing values natively by routing them to whichever child branch maximizes impurity gain, and handle categoricals natively via histogram integer binning and optimal Fisher splits.

Math & Tradeoffs→
M2-027Classical Machine Learning集成方法 (Bagging/RF)
Easy

Compare the Core Theoretical and Architectural Differences Between Bagging and Boosting

⚡ One-Sentence Answer

Bagging builds independent, deep trees in parallel on bootstrap samples to reduce variance; Boosting builds sequential, shallow trees iteratively on gradient residuals to reduce bias.

Math & Tradeoffs→
M2-028Classical Machine Learning集成方法 (Bagging/RF)
Easy

Explain the Two Sources of Randomness in Random Forests and How They Decorrelate Trees

⚡ One-Sentence Answer

Random Forests introduce randomness via (1) Bootstrap sample bagging (random rows with replacement) and (2) Feature subspace projection (random columns at each split); together they slash inter-tree correlation $\rho$ to minimize ensemble variance.

Math & Tradeoffs→
M2-029Classical Machine Learning集成方法 (Bagging/RF)
Medium

Detail Mean Decrease Impurity (MDI) vs. Permutation Importance (MDA) in Random Forests and Their Known Biases

⚡ One-Sentence Answer

Feature importance is measured via Mean Decrease Impurity (MDI, summing tree split gains) and Permutation Feature Importance (MDA, measuring score drops after shuffling a column); MDI is severely biased toward high-cardinality features, while MDA is robust.

Math & Tradeoffs→
M2-030Classical Machine Learning集成方法 (Bagging/RF)
Medium

Explain Out-of-Bag (OOB) Error Estimation and Its Practical Utility in Random Forests

⚡ One-Sentence Answer

OOB evaluation uses the $\approx 36.8\%$ of training observations omitted from each bootstrap tree to evaluate ensemble performance without requiring a separate validation set or cross-validation.

Math & Tradeoffs→
M2-031Classical Machine Learning集成方法 (Bagging/RF)
Hard

Define Stacking vs. Blending in Heterogeneous Ensembles and Compare Their Information Leakage Safeguards

⚡ One-Sentence Answer

Stacking trains a meta-learner on out-of-fold cross-validation predictions from multiple diverse base models, while Blending fits the meta-learner on a static holdout set.

Math & Tradeoffs→
M2-032Classical Machine Learning梯度提升 (GBDT/XGBoost)
Easy

Explain the Core Concept of GBDT: Why Sequential Trees Fit the Negative Gradient

⚡ One-Sentence Answer

GBDT performs gradient descent in function space; fitting each new tree to the negative gradient of the loss function $-\left[\frac{\partial L(y_i, F(x_i))}{\partial F(x_i)}\right]$ is the functional steepest descent step.

Math & Tradeoffs→
M2-033Classical Machine Learning梯度提升 (GBDT/XGBoost)
Medium

Survey the Core Algorithmic and Systems Innovations of XGBoost Over Traditional GBDT

⚡ One-Sentence Answer

XGBoost enhances GBDT via second-order Taylor expansion (Hessians), explicit leaf weight regularization (L1/L2), weighted quantile sketch for split proposals, column subsampling, and cache-aware hardware optimization.

Math & Tradeoffs→
M2-034Classical Machine Learning梯度提升 (GBDT/XGBoost)
Medium

Formulate the XGBoost Split Gain Equation and Explain Its Candidate Split Selection Mechanics

⚡ One-Sentence Answer

XGBoost evaluates candidate splits by computing the gain: $\text{Gain} = \frac{1}{2}\left[\frac{G_L^2}{H_L + \lambda} + \frac{G_R^2}{H_R + \lambda} - \frac{(G_L + G_R)^2}{H_L + H_R + \lambda}\right] - \gamma$; a split is accepted if and only if Gain exceeds zero.

Math & Tradeoffs→
M2-035Classical Machine Learning梯度提升 (GBDT/XGBoost)
Hard

Explain Prediction Shift in GBDT and How CatBoost Resolves It via Ordered Boosting and Target Statistics

⚡ One-Sentence Answer

CatBoost resolves prediction shift (target leakage in gradient and categorical encoding) by introducing Ordered Boosting and Ordered Target Statistics, which compute values strictly on permutations of past samples.

Math & Tradeoffs→
M2-036Classical Machine LearningSupport Vector Machines & Kernels
Easy

Explain the Maximum Margin Principle in Support Vector Machines and the Role of Support Vectors

⚡ One-Sentence Answer

SVM finds the unique linear hyperplane that maximizes the geometric margin to the nearest training instances; only the critical data points lying exactly on the margin boundary (Support Vectors) determine the decision boundary.

Math & Tradeoffs→
M2-037Classical Machine LearningSupport Vector Machines & Kernels
Easy

Formulate Soft-Margin SVM, Slack Variables, and the Regularization Role of Parameter C

⚡ One-Sentence Answer

Soft-margin SVM introduces slack variables $\xi_i \ge 0$ to tolerate non-separable margin violations, with hyperparameter $C$ governing the tradeoff between margin width and classification error penalties.

Math & Tradeoffs→
M2-038Classical Machine LearningSupport Vector Machines & Kernels
Medium

Explain the Kernel Trick, Mercer's Condition, and Implicit High-Dimensional Inner Products

⚡ One-Sentence Answer

The Kernel Trick computes the inner product in an infinite or high-dimensional feature space directly in the low-dimensional input space: $K(x, z) = \langle \phi(x), \phi(z) \rangle$, bypassing explicit non-linear feature expansion.

Math & Tradeoffs→
M2-039Classical Machine LearningSupport Vector Machines & Kernels
Medium

Compare Linear, Polynomial, and RBF Kernels and Their Decision Boundaries

⚡ One-Sentence Answer

Linear kernel is optimal for high-dimensional sparse data ($d \gg N$); RBF is the universal default for complex non-linear boundaries in moderate dimensions; Polynomial kernel is specialized for structured interactions (e.g. Computer vision/NLP degree-2 crossings).

Math & Tradeoffs→
M2-040Classical Machine LearningSupport Vector Machines & Kernels
Hard

Assess the Sensitivity of SVM to Class Imbalance and Label Noise and Industrial Mitigations

⚡ One-Sentence Answer

SVM is highly sensitive to class imbalance (which shifts the hyperplane toward the minority class) and label noise near the margin; mitigated via class-weighted penalties ($C_+ \ne C_-$) and slack variable bounds.

Math & Tradeoffs→
M2-041Classical Machine LearningNaive Bayes
Easy

Formulate Naive Bayes Classification Decision Rule and the Exact Nature of Its 'Naive' Assumption

⚡ One-Sentence Answer

The Naive Bayes decision rule is $\hat{y} = \arg\max_k P(Y=k) \prod_{j=1}^d P(X_j\mid Y=k)$; the 'naive' assumption posits that all features are mutually conditionally independent given the class label.

Math & Tradeoffs→
M2-042Classical Machine LearningNaive Bayes
Easy

Explain Laplace (Additive) Smoothing and Why It Prevents Zero-Probability Catastrophes in Naive Bayes

⚡ One-Sentence Answer

Laplace smoothing adds pseudo-counts $\alpha > 0$ to frequency estimators: $\hat{P}(x_j\mid y) = \frac{N_{jc} + \alpha}{N_c + \alpha V}$, preventing unobserved feature-class pairs from zeroing out the entire posterior probability.

Math & Tradeoffs→
M2-043Classical Machine LearningNaive Bayes
Medium

Compare Gaussian, Multinomial, and Bernoulli Naive Bayes Models and Their Respective Likelihoods

⚡ One-Sentence Answer

Gaussian Naive Bayes models continuous features via normal distributions; Multinomial Naive Bayes models word count frequencies; Bernoulli Naive Bayes models binary feature presence/absence while explicitly penalizing non-occurrences.

Math & Tradeoffs→
M2-044Classical Machine LearningNaive Bayes
Medium

Explain Why Naive Bayes Operates in the Log-Probability Domain to Prevent Underflow

⚡ One-Sentence Answer

Multiplying hundreds of probabilities $p_j \in (0, 1)$ triggers floating-point underflow to zero; converting to log-domain transforms products into sums: $\log P(Y) + \sum \log P(X_j\mid Y)$, ensuring numerical stability and faster execution.

Math & Tradeoffs→
M2-045Classical Machine LearningNaive Bayes
Hard

Contrast Naive Bayes and Logistic Regression as a Generative-Discriminative Model Pair (Ng & Jordan, 2001)

⚡ One-Sentence Answer

Naive Bayes (generative) models joint distribution $P(X, Y)$ and converges faster ($O(\log d)$ data); Logistic Regression (discriminative) directly models conditional $P(Y\mid X)$ and reaches a strictly lower asymptotic error ceiling ($O(d)$ data).

Math & Tradeoffs→
M2-046Classical Machine Learning梯度提升 (GBDT/XGBoost)
Medium

Compare XGBoost and LightGBM: Histogram Binning, GOSS, EFB, and Tree Growth Strategies

⚡ One-Sentence Answer

LightGBM outperforms traditional XGBoost in training speed and memory by introducing histogram-based continuous binning, Gradient-based One-Side Sampling (GOSS), Exclusive Feature Bundling (EFB), and leaf-wise (best-first) tree growth.

Math & Tradeoffs→
M2-047Classical Machine LearningK-Nearest Neighbors & Metric Learning
Easy

Describe the K-Nearest Neighbors (KNN) Algorithm Workflow and Computational Complexity

⚡ One-Sentence Answer

KNN is a non-parametric, lazy-learning algorithm that stores training data without explicit training ($O(1)$ train); inference computes distances to all $N$ training samples, selecting the $K$ closest neighbors to vote or average ($O(N d)$ test).

Math & Tradeoffs→
M2-048Classical Machine LearningK-Nearest Neighbors & Metric Learning
Easy

Explain How Hyperparameter K Governs the Bias-Variance Tradeoff in K-Nearest Neighbors

⚡ One-Sentence Answer

Small $K$ creates complex, fragmented decision boundaries resulting in low bias but high variance (overfitting); large $K$ smooths boundaries, resulting in low variance but high bias (underfitting).

Math & Tradeoffs→