Outlier detection methods for detecting cheaters in mobile gaming
summary
Mobile cheaters are flagged by measurable claims not topics: the pipeline uses single‑variable caps, Mahalanobis distance and minimum volume ellipsoid for multivariate outliers, and an autoencoder/ replicator neural net (relaxed step activations, early stopping, TensorFlow) to score errors. The uploader notes ~2% conversion so high spenders mimic cheaters; manual labeling and resampling by outlier score produces ground truth (one game yields eight actionable heuristics, another a trained classifier). Implementations use scikit‑learn and GitHub tools; false positives drive human review.
AI-written summary · about
chapters (15)
- 0:00Mobile gaming analytics intro
- 1:03Cheaters in social games
- 2:09Free-to-play revenue outliers
- 3:54Player view of cheating impact
- 4:38How players cheat
- 6:03Outlier detection vs cheating
- 7:07Single-variate outlier methods
- 8:08Multivariate Mahalanobis distance
- 10:17Robust ellipsoid methods
- 12:26Neural networks for outliers
- 17:01Variable selection for outliers
- 18:26Ground truth strategies
- 20:12Outlier detection in real games
- 21:41Limitations of outlier detection
- 23:06Q&A
related talks
30mCheaters’ Secrets Unveiled – How Cheaters Operate in Mobile Games & How to Counteract Them
29mThe Real Costs of Cheating on Studios, Mobile Games & Mobile Gamers Communities
18mOutsmarting the System: What Game Cheaters Can Teach Us About Cyber Security
36mFraud in mobile applications: how to define and detect
16mLeveraging ML to Create a Reliable Anti-Fraud System
39mSniffing Out the Strange — A Look at Anomaly Detection
30mMachine Learning Techniques for Class Imbalances & Adversaries
15mUsing Statistical Techniques to Automatically Detect Game-Breaking Issues