🏗️ System DesignID: imbalanced-cost-sensitive-learning

Imbalanced & Cost-Sensitive Learning

Q: Derive Focal Loss $\text{FL}(p_t) = -\alpha_t (1 - p_t)^\gamma \ln(p_t)$ and explain how the focusing parameter $\gamma$ downweights easy examples?

【Imbalanced & Cost-Sensitive Learning 核心考点 1】Derive Focal Loss $\text{FL}(p_t) = -\alpha_t (1 - p_t)^\gamma \ln(p_t)$ and explain how the focusing parameter $\gamma$ downweights easy examples?。技术解析：Definition: Imbalanced Classification & Cost-Sensitive Machine Learning addresses the extreme sample skew (fraud positive prevalence <0.01\% \sim 0.1\%, ratios exceeding 1:10,000) and highly asymmetric misclassification costs in financial risk modeling; it operates across 3 syner... 完整推导与工程落地见：https://talentme.airsota.com/resources/tech/roadmap/mindmap/System_Design/card/imbalanced-cost-sensitive-learning

Q: Compare EasyEnsemble vs BalanceCascade in training multiple bagging learners over partitioned negative sample majorities?

【Imbalanced & Cost-Sensitive Learning 核心考点 2】Compare EasyEnsemble vs BalanceCascade in training multiple bagging learners over partitioned negative sample majorities?。技术解析：Definition: Imbalanced Classification & Cost-Sensitive Machine Learning addresses the extreme sample skew (fraud positive prevalence <0.01\% \sim 0.1\%, ratios exceeding 1:10,000) and highly asymmetric misclassification costs in financial risk modeling; it operates across 3 syner... 完整推导与工程落地见：https://talentme.airsota.com/resources/tech/roadmap/mindmap/System_Design/card/imbalanced-cost-sensitive-learning

Q: Why is Accuracy prohibited in fraud evaluation, and how do KS statistic, PR-AUC, and Top-1% Capture Rate serve as authoritative metrics?

【Imbalanced & Cost-Sensitive Learning 核心考点 3】Why is Accuracy prohibited in fraud evaluation, and how do KS statistic, PR-AUC, and Top-1% Capture Rate serve as authoritative metrics?。技术解析：Definition: Imbalanced Classification & Cost-Sensitive Machine Learning addresses the extreme sample skew (fraud positive prevalence <0.01\% \sim 0.1\%, ratios exceeding 1:10,000) and highly asymmetric misclassification costs in financial risk modeling; it operates across 3 syner... 完整推导与工程落地见：https://talentme.airsota.com/resources/tech/roadmap/mindmap/System_Design/card/imbalanced-cost-sensitive-learning

Q: Derive the optimal Bayesian decision threshold $p^* = \frac{C_{\text{FP}}}{C_{\text{FP}} + C_{\text{FN}}}$ from cost matrix parameters $C_{\text{FP}}$ and $C_{\text{FN}}$?

【Imbalanced & Cost-Sensitive Learning 核心考点 4】Derive the optimal Bayesian decision threshold $p^* = \frac{C_{\text{FP}}}{C_{\text{FP}} + C_{\text{FN}}}$ from cost matrix parameters $C_{\text{FP}}$ and $C_{\text{FN}}$?。技术解析：Definition: Imbalanced Classification & Cost-Sensitive Machine Learning addresses the extreme sample skew (fraud positive prevalence <0.01\% \sim 0.1\%, ratios exceeding 1:10,000) and highly asymmetric misclassification costs in financial risk modeling; it operates across 3 syner... 完整推导与工程落地见：https://talentme.airsota.com/resources/tech/roadmap/mindmap/System_Design/card/imbalanced-cost-sensitive-learning

Q: How do unsupervised anomaly detection models (Isolation Forest, AutoEncoders) capture novel zero-day fraud without labeled positives?

【Imbalanced & Cost-Sensitive Learning 核心考点 5】How do unsupervised anomaly detection models (Isolation Forest, AutoEncoders) capture novel zero-day fraud without labeled positives?。技术解析：Definition: Imbalanced Classification & Cost-Sensitive Machine Learning addresses the extreme sample skew (fraud positive prevalence <0.01\% \sim 0.1\%, ratios exceeding 1:10,000) and highly asymmetric misclassification costs in financial risk modeling; it operates across 3 syner... 完整推导与工程落地见：https://talentme.airsota.com/resources/tech/roadmap/mindmap/System_Design/card/imbalanced-cost-sensitive-learning

极端不平衡样本与代价敏感学习

🎯Core Definition

Imbalanced Classification & Cost-Sensitive Machine Learning addresses the extreme sample skew (fraud positive prevalence

<0.01\% \sim 0.1\%

, ratios exceeding

1:10,000

) and highly asymmetric misclassification costs in financial risk modeling; it operates across 3 synergistic tiers: 1) Data Resampling (SMOTE oversampling and EasyEnsemble/BalanceCascade bagged negative downsampling); 2) Loss Re-weighting via Focal Loss

\text{FL}(p_t) = -\alpha_t (1 - p_t)^\gamma \ln(p_t)

(dynamically suppressing gradients from easy legitimate negatives to focus capacity on hard fraud cases); 3) Cost-Sensitive Decision Matrices that explicitly quantify the asymmetric business penalties of False Positives (user friction/churn) vs False Negatives (unrecoverable financial chargebacks) to calibrate optimal threshold cutoffs.

💡Use Cases

Payment chargeback prevention, AML transaction screening, loan default risk prediction, and malicious bot mitigation.

⚡Key Problems Solved

Standard binary cross-entropy collapses into predicting all transactions as legitimate (yielding 99.99% dummy accuracy while leaking all fraud); cost-sensitive learning minimizes monetary loss while maintaining strict false-positive caps.

🎯5 High-Frequency Exam Points

Derive Focal Loss

\text{FL}(p_t) = -\alpha_t (1 - p_t)^\gamma \ln(p_t)

and explain how the focusing parameter

\gamma

downweights easy examples?

Compare EasyEnsemble vs BalanceCascade in training multiple bagging learners over partitioned negative sample majorities?

Why is Accuracy prohibited in fraud evaluation, and how do KS statistic, PR-AUC, and Top-1% Capture Rate serve as authoritative metrics?

Derive the optimal Bayesian decision threshold

p^* = \frac{C_{\text{FP}}}{C_{\text{FP}} + C_{\text{FN}}}

from cost matrix parameters

C_{\text{FP}}

and

C_{\text{FN}}

How do unsupervised anomaly detection models (Isolation Forest, AutoEncoders) capture novel zero-day fraud without labeled positives?

📖 In-depth Guide:📄 risk-control-and-fraud-detection →

Updated 2026-08-14

🎯

Test Your Knowledge: Practice Questions for "Imbalanced & Cost-Sensitive Learning"

Single choice pitfall questions with instant feedback and mistake tracking.

🚀 Start Card Practice ➔

← Previous CardFlink Real-Time Feature Stream Next Card →Graph Risk & Collusion Detection

🔗 More System Design Knowledge Cards

RecSys Multi-Stage Funnel & 50ms SLA DSSM Two-Tower Retrieval YouTube DNN Candidate Generation Pre-Ranking Lightweight Architecture